PROLab Plus Manual

Published on June 2016 | Categories: Documents | Downloads: 66 | Comments: 0 | Views: 352
of x
Download PDF   Embed   Report

Comments

Content

ProLab Plus
Manual
The software package for
method interlaboratory tests
and proficiency tests

Version 2.14

Version: 2.14
For Windows NT, Windows 2000, Windows XP, Windows Vista and Windows 7
December 2011
quo data GmbH
Kaitzer Straße 135
D-01187 Dresden
Germany
phone: +49 351 402 88 67 0
fax:
+49 351 402 88 67 19
Email: [email protected]
www.quodata.de
Software development
Dipl.-Ing. Norbert Schick
Omri Teufert
Robert Walther
Daniel Rothmaler
Sebastian Falk
Dipl.-Phys. Christian Bläul
PD Dr. Steffen Uhlig
Manual
Dipl.-Math. Sabine Antoni
Kirsten Simon
Meike Schreitz
PD Dr. Steffen Uhlig

The software described in this manual must only be used or copied according to the terms
and conditions of the license agreement.
Copyright 1998-2011: PD Dr. Steffen Uhlig, Dresden

Contents
1
2

3
4

5
6

Introduction.................................................................................................................. 5
Preparation................................................................................................................... 7
2.1
System requirements ............................................................................................. 7
2.2
Installation.............................................................................................................. 7
2.3
Use of interlaboratory test data of older ProLab Plus versions ............................... 8
2.4
Hotline ................................................................................................................... 8
Overview ...................................................................................................................... 9
General information....................................................................................................12
4.1
Mouse functions ....................................................................................................12
4.2
Database handling ................................................................................................12
4.3
Important buttons ..................................................................................................12
4.4
Overview of buttons in main window .....................................................................13
Selecting ring tests.....................................................................................................14
5.1
Putting ring tests in edit mode ...............................................................................14
File menu.....................................................................................................................16
6.1
Settings in reports, tables and charts ....................................................................16
6.1.1
6.1.2
6.1.3

6.2
6.3
6.4
6.5
6.6
6.6.1
6.6.2
6.6.3

7

General settings ............................................................................................................... 16
Report settings ................................................................................................................. 17
Caption settings ............................................................................................................... 18

Start report ............................................................................................................19
Log info .................................................................................................................19
New database .......................................................................................................20
Change database..................................................................................................20
Backup file ............................................................................................................21
Restore from file............................................................................................................... 21
Store database to file ....................................................................................................... 21
Advice for long-term storage............................................................................................ 21

Database menu ...........................................................................................................22
7.1
Ring test selection.................................................................................................22
7.2
Basic tables ..........................................................................................................22
7.2.1
7.2.2
7.2.3
7.2.4
7.2.5

7.3
7.3.1
7.3.2
7.3.3

7.4
7.4.1
7.4.2

7.5
7.6
7.7

How to work with the tables ............................................................................................. 22
Ring tests ......................................................................................................................... 27
Samples ........................................................................................................................... 28
Measurands ..................................................................................................................... 29
Laboratories ..................................................................................................................... 30

Structure of ring test ..............................................................................................31
Laboratory allocation ....................................................................................................... 32
Sample-measurand allocation ......................................................................................... 33
Sample-measurand-laboratory allocation ........................................................................ 34

Encoding ...............................................................................................................37
Encoding laboratories ...................................................................................................... 37
Encode test portions ........................................................................................................ 38

Participating labs and data yet to be submitted .....................................................40
Characteristics of measurands ..............................................................................41
Test results ...........................................................................................................43

quo data GmbH

7.8
7.9
7.10
7.11

Creation of corrected measurands ........................................................................45
Analytical methods and sample preparations ........................................................46
Duplicate ring test .................................................................................................48
Export ...................................................................................................................48

7.11.1
7.11.2
7.11.3
7.11.4

7.12

Test results as CSV file ................................................................................................... 48
Laboratory files ................................................................................................................ 50
Ring test ........................................................................................................................... 54
Email merges ................................................................................................................... 55

Import ...................................................................................................................61

7.12.1 Import of structure with or without test results ................................................................. 61
7.12.2 Addresses, Measurands & samples, Analytical methods/Sample preparations ............. 64
7.12.3 Ring test ........................................................................................................................... 64

8

Test on homogeneity and stability ............................................................................65
8.1
Import test results .................................................................................................65
8.2
Encoding test portions...........................................................................................66
8.2.1 Uniform encoding of all test portions (for laboratories and for homogeneity analysis) and
random selection of homogeneity test portions ............................................................................ 66
8.2.2 Random selection of test portions for homogeneity test without encoding test portions for
laboratories ................................................................................................................................... 69
8.2.3 Another example for encoding test portions .................................................................... 70

8.3
8.4

Test results ...........................................................................................................71
Computation .........................................................................................................72

8.4.1
8.4.2
8.4.3

Assignment of test portions ............................................................................................. 72
Homogeneity .................................................................................................................... 72
Stability ............................................................................................................................ 76

9

Data selection .............................................................................................................79
9.1
Overview ...............................................................................................................79
9.2
Sample-measurand selection ................................................................................79
9.3
Sample-measurand-laboratory selection ...............................................................81
9.4
Save and load selection ........................................................................................83
10 Statistics .....................................................................................................................84
10.1 Overview ...............................................................................................................84
10.2 Ring test parameters .............................................................................................84
10.2.1
10.2.2
10.2.3
10.2.4

10.3

Option for DIN 38402 A 45: logarithmic calculation ......................................................... 87
Determine assigned value ............................................................................................... 88
Determine target s.d. ....................................................................................................... 90
Variance function ............................................................................................................. 91

Computation of Z-scores .......................................................................................92

10.3.1 Computation tab .............................................................................................................. 92
10.3.2 Score values tab .............................................................................................................. 95
10.3.3 Results tab ....................................................................................................................... 96

11 Charts and tables........................................................................................................98
11.1 Summary results ...................................................................................................99
11.1.1 Right side bar ................................................................................................................. 101
11.1.2 Menu Format.................................................................................................................. 102
11.1.3 Menu Output .................................................................................................................. 102

11.2
11.3

Survey of scores .................................................................................................104
Combination scores ............................................................................................106

11.3.1 Chart of RLP and RSZ ................................................................................................... 106
11.3.2 Laboratory assessment based on RLP ......................................................................... 112
quo data GmbH

11.3.3 Distribution of Z scores .................................................................................................. 118
11.3.4 Control chart and extended evaluation .......................................................................... 123

11.4

Test on equivalence ............................................................................................123

11.4.1
11.4.2
11.4.3
11.4.4
11.4.5
11.4.6
11.4.7
11.4.8

Introduction .................................................................................................................... 123
Define classes of methods ............................................................................................. 124
Recovery equivalence ................................................................................................... 126
Reproducibility equivalence ........................................................................................... 127
Repeatability equivalence .............................................................................................. 128
Sample specific method effects ..................................................................................... 128
Survey of methods ......................................................................................................... 129
Export of results ............................................................................................................. 129

11.5 Tolerance limits and rel. standard deviations ......................................................129
11.6 Youden plot.........................................................................................................130
11.7 Histogram and normal plot ..................................................................................133
11.8 Chart of lab means and repeatability standard deviation .....................................136
11.9 Kernel density estimator (KDE) ...........................................................................138
11.10 HORRAT trend....................................................................................................141
11.11 Principal Component Analysis (PCA) ..................................................................143
11.12 Mandel’s h & k statistics ......................................................................................145
11.13 Laboratory mean values......................................................................................148
11.14 Further reports ....................................................................................................149
12 Quick start: The steps of statistical analysis ..........................................................151
13 Some aspects of the statistical methods in ProLab Plus ......................................156
13.1 When can the method of the Swiss food manual be applied? .............................156
13.2 When can Q/median, Q/Huber or DIN 38402 A45 be applied?............................156
13.3 When can ISO 5725 and DIN 38402 A42 be applied? ........................................156
13.4 When can the Horwitz function be applied? ........................................................156
13.5 When can the empirical Horwitz function be applied? .........................................157
13.6 When can the nested design analysis according to ISO 5725-3 and ISO 5725-5 be
applied? ..........................................................................................................................157
14 References ................................................................................................................158

quo data GmbH

quo data GmbH

.

ProLab Plus – Manual Introduction

5

1 Introduction
ProLab Plus is a software package specially conceived to facilitate a professional and efficient performance of the various tasks which arise in the context of interlaboratory tests, from
statistical analysis to professional data handling. ProLab Plus can be used for the certification of reference materials, the evaluation of analytical methods and for the assessment of
laboratories.
ProLab Plus will prove useful at every step of a proficiency test. It provides tools which will
assist you right from the planning stage, to the realization of homogeneity tests, the automatic data input from laboratories, the application of statistical analysis and the generation of reports.
Several statistical methods are available:
-

-

-

ISO 5725-2
DIN 38402 A42 (German standard for
method ring tests for water analyses)
Swiss Food Manual
Q-Method/Median (Robust statistical
method)
Q-Method/Huber (Robust method for
proficiency tests used e.g. by the German co-operation for water analyses
‘LAWA’)

-

-

DIN 38402 A45 (Robust Q-method and
Hampel estimator)
ISO 5725-5 (algorithm A+S)
Nested Design 2x2 (ISO 5725-3/5)
Nested Design (ISO 5725-3)
(Empirical) Horwitz function
ISO 13528

As for the assessment of laboratories, there are also several available protocols, such as the
international Harmonized Protocol of IUPAC, ISO and AOAC1, ISO 13528, DIN 38402 A45,
CD ISO 20612 and the LAWA2 protocol. More specifically, the laboratories can be assessed
by
-

Z scores, Zu scores, Zeta scores, Z’ scores
Youden plots
RSZ (Rescaled Sum of Z Scores)
RLP (Relative Laboratory Performance)
Percentage of successful measurements

ProLab Plus allows flexible data input, whether manual or through the import of digital files.
The simple data entry program RingDat is provided which may be forwarded to the participating laboratories without extra charge. Additional information on the laboratories and the
methods can be entered into the system, such as addresses, the determination limit and the
detection limit, the sample preparation method and the analytical method.
1

Michael Thompson and Roger Wood (1993), International Harmonized Protocol for Proficiency Testing of
(Chemical) Analytical Laboratories. Journal of AOAC International 1993, Vol. 76, 926-940.
2
AQS der Länderarbeitsgemeinschaft Wasser für Wasser-, Abwasser- und Schlammuntersuchungen (Working
Group on water issues of the federal states and the federal government)

quo data GmbH

.

ProLab Plus – Manual Introduction

6

A powerful data selection program assists the user in the evaluation and the presentation of
results, enabling him to sift data as convenient, to generate individual reports or reports for
selections of laboratories, and to set up and save selection patterns. Up to four selection patterns per ring test can be saved.
All results are stored in the database. Comments can be saved for participants, measurements and within charts.
Numerous specialised chart-generating functions are provided to assist in the presentation of
results, while a powerful report designer enables the generation of editable reports. Reports
can be exported into the formats PDF, RTF (Word), XLS (Excel), HTML, Text and CSV.
Charts can be converted into Bitmaps (BMP), Scalable Vector Graphics (SVG) and the Windows Meta File format (WMF).

quo data GmbH

.

ProLab Plus – Manual Preparation

7

2 Preparation
2.1

System requirements

The hardware and software requirements for ProLab Plus are listed in the following table.
Processor
RAM
Hard disk
Operating System

Intel Pentium or compatible with at least 1 GHz (> 2.4 GHz recommended)
512 MB
> 300 MB free
Windows NT (with service-pack 5), Windows 2000, XP Professional
(not Windows XP Home Edition), Windows Vista and Windows 7.

ProLab Plus runs stable under Windows 7, and all calculations and reports work as expected.

2.2

Installation

1. If you have an existing ProLab Plus version, please backup your current ProLab Plus
program directory and your database before installing the new version into the current
program directory. The path to the database is shown at the very top of the ProLab Plus
window.
2. If you downloaded a setup program from our website, please run it now. If you want to install ProLab using a CD, please put the CD into the drive. The setup program will start
automatically. If not, please run setup.exe in the root directory of the CD.
3. Follow the installation dialogue and use the directories proposed. Upon successful installation, a program group ProLab will be created, which contains ProLab, a help file for the
Report Designer (Qrduser) and a program for direct input of data from laboratories
(RingDat).
4. Run ProLab Plus.
5. After the first start you will see a registration dialogue with a registration number. You can
phone quo data GmbH (+49 351 402 88 67 0) or send an email ([email protected]) and
you will get the necessary key. After input of the key press “Save”.
Alternatively, you can proceed as follows. Close the program ProLab Plus. In the program
directory you will find the file REGISTER.INI. Send this file via email to [email protected].
Within one working day, you will receive a new file REGISTER.INI with the key. This file
must be copied into the program directory of ProLab, replacing the original file.

quo data GmbH

8

2.3

.

ProLab Plus - Manual Preparation

Use of interlaboratory test data of older ProLab Plus versions

Since ProLab 2003, a new database format is used (ADS, developed by Sybase which belongs to SAP). The current version ProLab Plus 2.14 uses ADS 10. It is therefore not possible to directly access older BDE databases. However, the conversion program
„ProLabMigrator.exe" allows simple conversion of the old databases. Start this program, select the directories for the BDE database and the ADS database and start the transfer.

2.4

Hotline

In case of problems use the hotline of quo data: +49 351 402 88 67 0 or send an email to:
[email protected].

quo data GmbH

.

ProLab Plus – Manual Overview

9

3 Overview
The main menu is the starting point of every evaluation using ProLab Plus. You are in the
menu as soon as you successfully start ProLab Plus.
The main menu contains eight modules, which can be selected using the common Windows
techniques (mouse, arrow keys or input of the corresponding character):
-

File
Database
Homogeneity / Stability
Selection
Statistics
Charts and tables
Window
Help

The modules are ordered in such a way that, in the performance and evaluation of a ring
test, they are typically activated from left to right. One starts in the module File with the definition of settings for reports and data base tables. That is followed by the description of the ring
test, of the samples and measurands that are to be examined and evaluated, and of the encoding of laboratories and samples. Once the results start coming in, they will thus be able to
be imported or manually entered. All necessary functions for the latter procedure are located
in the module Database. If required, a test of the homogeneity of the samples in preparation
for the ring test can be conducted in the module Homogeneity / Stability, which also provides
the necessary functions for any stability tests to be conducted later on. The Selection module
provides all relevant tools for the selection of samples, measurands and laboratories to be
included in assessment and analysis procedures or in the presentation of results. The evaluation, analysis and assessment procedures are then carried out in the module Statistics,
while reports and charts in various layouts can be generated and displayed in the module
Charts and tables. Each step within this sequence affects subsequent steps. All statistical results obtained at any stage will be directly stored in the database. Earlier results of the same
ring test will be overwritten automatically. If there are some modifications to implement ( e.g.
in the calculation of Z scores) it is not necessary to create a new test: changing the data at
the earliest relevant step is sufficient.
An overview of important functions in ProLab Plus is given in the following table.

quo data GmbH

.

ProLab Plus – Manual Overview

10

File

Settings
Formats, general report settings and report designer
Generate database, change database, backup files

Database

Database and data input
Ring test selection
Basic tables: ring tests, samples, measurands (analytes), laboratories
Structure of ring test: assignment of laboratories and measurands
Encoding: Encoding of laboratories and samples
Participating labs and outstanding data
Characteristics of measurands: reference values, units, conversion factors et al.
Definition of analytical methods and sample preparations
Test results
Creation of corrected measurands
Duplicate ring test
Export: CSV files, laboratory files, ring test, email merge
Import: structure/test results, addresses, measurands & samples, analytical
methods & preparations and ring tests

Homogeneity /
Stability

Import and analysis of homogeneity and stability test results
Import test results
Test portion encoding
Test results
Computation of homogeneity and stability

Selection

Selection of data for statistical analysis and the presentation of results
Selection of samples and measurands
Selection of samples, measurands and laboratories
Save and load selections

Statistics

Analysis and assessment
Calculation of statistical parameters of the ring test (lab means, lab s.d., overall
mean, reproducibility s.d. and repeatability s.d., specification of target value and
target s.d.)
Calculation of tolerance limits and single scores (Z scores, Z u scores, Zeta
scores, Z’ scores)

Charts and

Charts and tables of the statistical analysis

Tables

Comprehensive presentation of results for each sample-measurand combination
separately
Survey of scores
Combination scores and ring test-spanning analyses
Test on equivalence
Tolerance limits and rel. standard deviations

quo data GmbH

.

ProLab Plus – Manual Overview

11

Youden plot
Histogram and normal plot
Chart of repeatability standard deviations
Kernel density estimator (KDE)
HORRAT trend
Principal component analysis
Mandel’s h & k statistics
Laboratory mean values per sample and per measurand
Further reports: number of test results and labs, survey of test results, method
characteristics, summary of outliers, individual reports
Help

Help function
Program update: check for new version

quo data GmbH

.

ProLab Plus – Manual General hints

12

4 General information
4.1

Mouse functions

The main functions of the program can be activated by using the left mouse button. Some
additional functions ( e.g. number of columns and column-width in tables) can be activated
by using the right mouse button.

4.2

Database handling

In some forms, data sets ( e.g. measurands, ring tests, labs, samples, test results) can be
accessed, edited, added or deleted. The following commands are available for this purpose:
Back

Go to the previous record

Forward

Go to the next record

Insert

Insert new record

Delete

Delete selected record

Save

Save changes

Cancel

Cancel editing

Please note that it is not possible to undo changes in the database. Therefore it is very important to make a backup of the database at the beginning of each session, so the original
database can be recovered. For the creation of the backup select File – Backup file.

4.3

Important buttons

The following list explains some important buttons frequently used in the program:
Close
Closes current window.
Report

Click this button to display a report preview for the selected data. In many
cases, it is also possible, by simultaneously holding down the SHIFT key
, to open a spreadsheet which allows data editing and storage in a way
similar to the Excel program.

Help

Activates help for the current window.

Save

Click this button to save tables or charts as a file.

Print

Click this button to print out reports or charts.

Lab code

Click this button to turn switch between the laboratory name (LName) and
the laboratory code (LCode) in charts and tables.

quo data GmbH

.

ProLab Plus – Manual General hints

4.4

13

Overview of buttons in main window

The buttons in the main-menu toolbar make it possible to easily activate the most important
and commonly used functions. The following table offers an overview of these shortcuts.
Change ring test and define working mode
Settings
Survey of test results
Edit data of basic tables
Assignment of participating laboratories
Assignment of measurands to samples
Assignment of measurand-sample combinations to laboratories
Encoding laboratories
Encoding test portions
Ring test-specific characteristics of measurands
Edit test results
Select sample / measurand for further use
Select sample / measurand / laboratory for further use
Save selection
Load previously saved selection
Calculation of ring test parameters
Calculation of scores and laboratory evaluation
Chart/table of laboratory mean values and standard deviations
Survey of scores
Chart of tolerance limits and rel. standard deviations
Youden plot
Histogram und normal plot
Table of laboratory mean values and scores – per measurand
Table of laboratory mean values and scores – per sample
Help

quo data GmbH

.

ProLab Plus – Manual Selecting ring tests

14

5 Selecting ring tests
Many features of ProLab Plus are specific to the selected ring test. Subsequently used modules and functions always refer to the ring test previously selected in the selection box.

Figure 1: Main menu

There are two different working modes for a ring test. They can be selected from the main
menu via the button Ring test or via the menu Database – Ring test selection.

5.1

Putting ring tests in edit mode

If you want to work on a ring test, e.g. enter test results, carry out calculations etc., select
the checkbox in the Access column for the relevant ring test and double click on the ring test,
or close the window and select it from the selection box on the top of the main window.

Figure 2: Database – Ring test selection

quo data GmbH

.

ProLab Plus – Manual Selecting ring tests

15

If you only want to view the results and prepare presentations, do not select the check box,
but double click on the relevant ring test or select it from the selection box on the top of the
main window. No data input or changes are possible in this mode.
If all ring tests should be included in a presentation, all of them need to be made accessible.
Therefore click on the button Make all ring tests accessible. If you don’t want any of the ring
tests in the edit mode, click on Make all ring tests read-only.
If a ring test is marked in red, it is currently opened on a different computer. You can also
open it and make changes, but this may lead to inconsistencies in the database.

quo data GmbH

.

ProLab Plus – Manual File Menu
16

6 File menu
The File menu contains settings for reports and tables, the report designer and selection of
the database.

6.1

Settings in reports, tables and charts

The Settings dialogue box allows you to choose different settings for how information will be
displayed in the reports, tables and charts. This includes the specification of number formats
and the name of the institution. Furthermore, the currently used logo is shown. It may be
changed via the button Change.
Changes will be saved automatically upon closing the window or switching between tabs.
6.1.1 General settings
This tab enables you to enter general settings such as institution name, number format, creation date, chart size and colours. If the check boxes for number format, date or default chart
size are marked, these settings will apply to all tables, forms and graphics displayed.

Figure 3: File – Settings – General

The characters for formatting numerical values are:

.

Decimal separator

0

this digit must be a numeric character

#

this digit may be a numerical character

E+

exponential notation

quo data GmbH

.

ProLab Plus – Manual File Menu

17

By clicking on Reset chart settings all saved chart settings and profiles as well as information
regarding size and colour are reset to the default setting. By clicking on Reset non-chart settings and confirming the list of configuration files to be deleted, all information regarding institution, date and number format and many other settings stored by other ProLab windows are
reset to the default setting.
By marking the checkbox Highlight outliers of type E in the summary results this option can
be selected. For further explanation of outliers of type E please refer to chapters 10.2 and
11.1.
6.1.2 Report settings
The tab Reports contains a comprehensive list of commonly used report templates. ProLab
Plus offers a wide variety of templates, so the time-consuming task of designing reports can
thereby often be avoided.

Figure 4: File – Settings – Reports

Select a report and click this button to open the Report designer module. Alternatively, a double click on the report will open the same module. Any changes become ef-

quo data GmbH

.

ProLab Plus – Manual File Menu

18

fective upon saving the report in the Report Designer module. Customised report are
highlighted in bold writing.
Since for some reports the data is missing in the templates, the report might look
somewhat strange at first when using the preview. Upon input of data, however, you
will find that the report looks fine.
Using this button will reproduce the original report template as created by the installation program. This button is only relevant for reports highlighted in bold writing, that is
to say, for customised report templates.
Designing new reports is possible, either directly or by using the Report Designer module.
This button opens the Report Designer for creating a new report.
This button allows you to edit existing reports.
User-defined reports can be accessed via the menu Charts and tables – Further reports –
Own reports.
The last column in the table shows the date of the last adjustment of the report template, not
the date of the last time the report was used.
6.1.3 Caption settings
The Captions tab contains standard abbreviations for a number of names for titles and headers in charts and tables (e.g. column headers) and enables you to customise them. Please
be aware that these settings will not become effective in every report. In some cases, the
particular report has to be adjusted using the Report Designer.

quo data GmbH

.

ProLab Plus – Manual File Menu

19

Figure 5: File – Settings – Captions

6.2

Start report

By choosing the menu item File – Start report or pressing F8, you can determine the page
number on which the next report you want to print will start. This is helpful when there are
several reports to be compiled, i. e. you want to combine them into one report with successive page numbers.

6.3

Log info

During a ProLab Plus session the following information is saved:
Time of log in and log out
User name
Imported data
Test results that have been changed
Protocol of calculation of ring test parameters
Protocol of calculation of scores.

quo data GmbH

.

ProLab Plus – Manual File Menu

20

To access this information please choose Log info from the File menu. To open the report
preview containing the information from all ProLab Plus sessions use the Report button.

Figure 6: File – Log info

6.4

New database

Via File – New database a new and empty database can be generated automatically in a
new directory. This might be appropriate when different ring tests should be saved in different databases. Generally, saving ring tests in one and the same database might be reasonable, e.g. for an analysis spanning several ring tests; however, too many ring tests in the
same database may slow down the program noticeably.
After selecting a directory from the list or entering a name for a new directory in the edit window, the structures of the tables in the current database are copied to the target directory.
WARNING: Existing tables in the target directory will be overwritten whereby any data contained in these tables will be lost.

6.5

Change database

It is possible to change database directories, that is to say, to transfer a database with all the
data relative to a ring test from one directory to another. Make sure that the new database directory contains all required tables, otherwise your work may be compromised. This is best
achieved by using the Backup file command or by copying the database directory outside of

quo data GmbH

.

ProLab Plus – Manual File Menu

21

ProLab Plus.

6.6

Backup file

This command allows backing up the entire database (all tables) in a compressed file and
recovering it again from this file if necessary. LHARC is used as compression format. For
saving your data, select a target folder and enter a file name.
When quitting ProLab Plus you will be asked whether you want to create a backup copy of
your database. Additionally, you can manually backup the database to a folder of your
choosing by using the following tab.

Figure 7: File – Backup file (Backup of database „Demo“)

6.6.1 Restore from file
In this window, all backup files that have been created so far (and still exist) are listed. In order to restore one of them, simply select it from the list and click the button Restore from file.
Alternatively you can choose Select file… in the first row of the table and specify the LZH file
you want to restore.
WARNING: Any existing data will be overwritten and thereby lost.
If you want to restore a database without overwriting existing data and tables, please create
a new database as described in section 6.4 before restoring.
6.6.2 Store database to file
In order to create a backup of your database, you have to select a target file first. Normally
that will be a new file, which can be created via the first row Select file… Alternatively an existing file from the list can be chosen. To start the process simply click Store database to file.
Please note that any existing data will be overwritten and thereby lost.
6.6.3 Advice for long-term storage
Although there are several programs of other developers or Open Source communities which
can open LZH files and/or the contained Advantage Database System files, this binary format may not be the best choice for long-term storage of ring test data. It is recommended instead to export plain-text CSV files additionally (please refer to chapter 7.11.1).

quo data GmbH

.

ProLab Plus – Manual Menu Database

22

7 Database menu
7.1

Ring test selection

Please refer to chapter 5.1, Putting ring tests in edit mode.

7.2

Basic tables

A ring test is defined by the samples, the measurands and the laboratories. Therefore, the
tables for the ring tests, samples, measurands and laboratories constitute the foundation of
the database. The basic tables can be reached via Database – Basic tables. In these tables,
all the information on samples, measurands and laboratories is entered, but no allocation (
e.g. e.g. which measurand is analysed by which laboratory on which sample) is carried out
yet.
7.2.1 How to work with the tables
The data tables offer a powerful tool for working with the database. The data can be sorted
and grouped in several ways, fields can be added and removed and whole tables or selections can be saved in different formats.
In the following, several possibilities for working with the tables are described. However, it
must be noted that not all commands are available in all tables.
The most important buttons are
Save

Save changes

Cancel

Cancel editing

Add

Add new data

Delete

Delete selected data

New data ( e.g. ring tests or samples) can also be added by clicking into the first row of the
respective table.
Please note that the names of the ring tests, the measurands, the samples and the laboratories are unique identification labels. Therefore they should not be changed later on. Name
must only contain capital letters and numeric characters. Description can be filled and/or
changed later. If the Description field is left empty, Name will be entered automatically.
When clicking with the right mouse button onto the column headers, a pop-up menu opens
within which the available commands are in dark letters (see Figure 8). These commands will
be explained in more detail in the following:

Reset table settings
All changes in the appearance of the table can be undone by choosing Reset table settings
from the pop-up menu. You get there by a right click with the mouse onto any column header.

quo data GmbH

.

ProLab Plus – Manual Menu Database

23

Sorting
The tables can be sorted in ascending or descending order (with respect to a selected column) either
by left-clicking in the respective column header or
by selecting the commands from the following pop-up menu, which opens by rightclicking into a column header.

Figure 8: Pop-up menu after right mouse button click into column header

Add and remove columns, adjust order
The order of the columns can be changed by drag & drop: simply click in a column header
and by keeping the mouse button pressed move the column to the desired position.
Unnecessary columns may be hidden by deselecting the respective column from the list that
can be reached via the button Visible columns. A column may also be removed by clicking in
the header and selecting Remove This Column from the pop-up menu. If formerly removed
columns or further columns should be added to the table, simply select the column from the
list under Visible columns or select the Field Chooser from the pop-up menu. From the field
chooser, a column can be moved to the table by drag and drop.
In order to add or remove several columns at once, right-click anywhere inside the table (not
in the headers) and select Select columns from the pop-up menu while keeping the Shift key
 pressed.
The column width can be adapted manually by moving the separators of a column. The best
fit, i. e. the column width that enables to see the whole content of the column without wasting
too much space, can be obtained by double clicking on a separator or by choosing Best Fit
from the pop-up menu. An optimal width of all columns can be obtained via the command
Best Fit (all columns) from the pop-up menu.
Grouping
A table can be grouped with respect to column entries and will thereby be divided into subtables. Grouping can be done by marking the desired column and selecting Group By This

quo data GmbH

.

ProLab Plus – Manual Menu Database

24

Field from the pop-up menu. After selecting this command, a dark grey box will appear above
the table indicating the column according to which the grouping is carried out. A further possibility for grouping is to select the command Group By Box from the pop-up menu. The dark
grey box then appears without any column header name. Now the column according to
which the grouping should be carried out can be moved to the dark grey box via drag and
drop.
In , the table grouping of measurands with regard to the column Category 2 is shown. As
there are three entries in the Category 2 column, the grouped table now consists of three
sub-tables, one for light metals, one for heavy metals and one for gasses. These sub-tables
can be opened or closed via the + / - buttons.
Of course, a table can be grouped several times in succession. Thereby more and more subtables are generated. This grouping can be done again either by successively marking column headers and selecting the command Group By This Field from the pop-up menu or by
selecting the command Group By Box once and then successively dragging fields to the grey
box above the table.
In both cases, the hierarchy of the grouping will be shown in the dark grey box.

Figure 9: Basic table of measurands grouped by Category 2

Column width
Using the menu option Best Fit, the width of the columns will be adjusted according to the
content. To adjust all columns at once the shortcut Ctrl + Addition (on the numerical keyboard) can be used.

quo data GmbH

.

ProLab Plus – Manual Menu Database

25

Copy
Whole tables or selection thereof can be copied to the clipboard and, e.g., inserted into an
Excel table. For this purpose use the right mouse button within the table (not in the headers).
Then a pop-up menu will be displayed where you can select Copy table or Copy selection.
With the first command, the whole table is copied. With the second command, only the row in
which has been clicked is copied, along with the column header.

Figure 10: Pop-up menu after right mouse button click into table

Save
All tables can be saved in different formats: text file, Excel file, XML file or HTML file. The table can be saved as a PDF or RTF file from within the report preview. For this purpose use
the right mouse button within the table (not in the headers). Then a pop-up menu will be displayed where you can select the desired saving format.
The information of a grouped table is always completely saved independently of the way it
has been displayed in ProLab Plus, i. e. the information of sub-tables closed in ProLab Plus
will be saved as well. Of course, the grouping is saved, too.

Search text
After clicking this menu item, a search bar will be shown. That bar can be opened by using
the shortcut Ctrl + F alternatively. To close the bar, click on the cross on the left side or use
the Esc key.
Select columns
This option opens a list with all columns that can be shown. There are three different types of
lists:
1. In most windows, the default setting leads to a list with check boxes, which makes
possible a swift selection of columns to show or hide.
2. If you decide to change the visibility of more than one column at once, hold the Shift
key down while clicking on the menu option ‘Select columns’.. A window appears in
which you can show or hide all columns by using the keys ‘>>’ or ‘<<’. It is also possi-

quo data GmbH

.

ProLab Plus – Manual Menu Database

26

ble to show or hide single columns in this window. Some tables only offer this window.
The third window can be reached by a right mouse click into a column header and selecting Field Chooser in the opening pop-up menu. The usage of that window is described above.

quo data GmbH

.

ProLab Plus – Manual Menu Database

27

7.2.2 Ring tests
The parent data for ring tests can be edited (entered, changed, deleted) in the Database –
Basic Tables dialogue box on the Ring tests page. The list comprises all open ring tests, i. e.
all ring tests that are opened in the Edit mode (see section 5).

Figure 11: Database – Basic Tables – Ring tests

Name stands for the abbreviated name of the respective ring test. In the database it is usually referred to as RName. To add a new ring test, either click in the box Add ring test or click
on . Enter the name of the ring test (Name) and, if necessary, the Description, the Date of
analysis, Report date, Cut-off date (unofficial deadline), Design and the number of Replicates
in the respective cells of the table. Name may only contain up to eight capital letters and numerics.
For every ring test, a design can be selected from the scroll-down box, though this is not
mandatory:
Select single measurement if only one test result will be reported for every laboratorysample-measurand combination.
Select with replicates and enter the number of replicate measurement values for every
laboratory-sample-measurand combination in the pop-up window. The number of replicates can be changed later if necessary. This will be used as default setting for the export of laboratory files (7.11.2) and in the window for the input of test results. The computation of means and standard deviations however is not affected by that setting, as
the number of actually available measurements is independently determined for these
functions..
Select Nested design if you would like to work with a ring test design where e.g. two
measurement values belong to one day and two values belong to another day (see
ISO 5725-3).
The selection of the design will affect the window for entering the test results, however the
design can always be changed again later, if need be. If no design is selected, the number of
replicates can be defined later on, in the test results window.
Further information (e.g. year, organizer) regarding the ring test can be entered in the field
Group. This may be helpful for a later grouping of the ring tests.

quo data GmbH

.

ProLab Plus – Manual Menu Database

28

Warning: If you delete a ring test, please note that this operation will also delete any data in
other tables that refer to this ring test ( e.g. samples, ring test-sample-measurand combinations, test results, lab codes). It does not delete measurand and laboratory data in the basic
tables.
7.2.3 Samples
To enter a new sample or change information regarding existing samples, please select the
tab Samples in the dialogue box Database – Basic Tables.
After selecting the corresponding ring test, a sample can be added either by clicking in the
box Add sample or on the
button. Afterwards Name and Description can be entered in the
corresponding fields. The name of the sample (in the database referred to as SName or
PName) is a unique identification, therefore it should not be changed later on. Name must
only contain capital letters and numerics.

Figure 12: Database – Basic Tables – Samples

For an easier selection of samples in the calculations, two categories are available under
which distinguishing features of the samples can be entered. The button Sorting opens a
new window, where the samples can be sorted by drag & drop or according to defined fields.
Please be aware that changes can only be made if the ring test is opened in the Edit mode.
Entering a new sample requires the existence of a ring test, as samples (other than measurands or laboratories) are always subordinated to a ring test.

quo data GmbH

.

ProLab Plus – Manual Menu Database

29

Warning: Please note that deleting a sample will also delete any data in other tables that refer to this sample ( e.g. ring test-sample-measurand combinations, test results, test portion
codes). Deleting a sample does not however delete ring tests, measurands or laboratory data in the basic tables.
7.2.4 Measurands
The parent data for measured measurands can be edited (entered, changed, deleted) in the
Database – Basic Tables dialogue box on the Measurands tab. Changes are only possible, if
the corresponding ring test is opened in the Edit mode. If the ring test is opened in the Read
mode, the measurand data can be read, but not changed.
To add a measurand, either click in the box Add measurand or click on . The table of
measurands is independent from the ring test and the sample. The necessary allocation will
be carried out later.
In the form you can enter the name of the measurand (Name – in the database referred to as
MName), the description of the measurand and – if desired – further distinguishing features
under two categories and the number format for the test results. The restriction of using only
capital letters and numerics for Name applies here as well.
If you want to change the order in which measurands appear elsewhere in ProLab, please
click the Sorting button. A pop-up window will appear where you can choose how the measurands should be sorted: by name, description or category (or manually via drag and drop).

Figure 13: Database – Basic Tables – Measurands

Warning: Please note that deleting a measurand will also delete any data in other tables that
refer to this measurand ( e.g. ring test-sample-measurand combinations, test results). Deleting a measurand does not however delete ring tests, samples or laboratory data in the basic
tables.

quo data GmbH

.

ProLab Plus – Manual Menu Database

30

7.2.5 Laboratories
The parent data for laboratories can be edited (entered, changed, deleted) in the Database –
Basic Tables dialogue box on the tab Labs as Table. To add a laboratory click in the box Add
laboratory or on the button in the lower part of the window.
The table of laboratories is independent from the ring test and the sample. The necessary allocation will be carried out later. Changes are only possible if the respective ring test is
opened in the Edit mode. If the ring test is opened in the Read mode, the laboratory data can
be read, but not changed.
You can enter the name of the laboratory, a description, the address and – if desired – further information such as contact person, email address, telephone numbers etc. Email addresses and contact persons are of special importance if you want to use the email merge
module of ProLab Plus (see also section 7.11.4). More than one address can be specified by
using spaces or semicolons as delimiters. Furthermore, up to two categories can be specified for each laboratory that allows for a later selection of sub-groups of laboratories.
Please note, that for Name only capital letters and numerics are allowed. If the Description
field is left empty or contains only space characters, the laboratory name will be entered automatically, since different reports and graphics use the description to specify the laboratory.
To show a tabular summary of all laboratories click the Report button.

Figure 14: Database – Basic Tables – Labs as Table

Displaying all information on a laboratory in a table often leads to a confusingly high number
of columns that can hardly be displayed at once. Therefore, on the tab Labs as Cards there
is also the option to show all laboratory information as business cards.

quo data GmbH

.

ProLab Plus – Manual Menu Database

31

Figure 15: Database – Basic Tables – Labs as Cards

As the address structure varies from one country to another, it is possible to enter all address
information in the additional fields Additional 1 to Additional 6 instead of the fields Street, ZIP
and City. This offers the possibility to create, for example, address labels with the correctly
arranged address. Thus (see also Table 1), for a German address the additional name of the
laboratory could be entered in the field Additional 1, the street with number in Additional 2
and the postal code and city in Additional 3. For the UK the additional name of the laboratory
could be entered again in the field Additional 1, the street with number in Additional 2, the
city in Additional 3, the county in Additional 4 and the postal code in Additional 5.
Germany

USA

UK

short name of lab

short name of lab

short name of lab

1 additional name

1 additional name

1 additional name

2 Street + No

2 No + Street

2 No + Street

3 ZIP City

3 City State/ZIP

3 City

Country

Country

4 County
5 ZIP
Country

Table 1: Examples of addresses

Warning: You may delete laboratories, but please note that this operation will also delete any
data in other tables that refer to this laboratory ( e.g. ring test-sample-measurand combinations, test results, lab codes). Deleting laboratories does not however delete ring tests, samples or measurands in the basic tables.

7.3

Structure of ring test

After the basic tables have been filled, it still has to be determined which samples were analysed by which laboratories with respect to which measurands. Hence it is necessary to as-

quo data GmbH

.

ProLab Plus – Manual Menu Database

32

sign the investigated measurands to the samples and the laboratories to the respective sample-measurand combinations.
Therefore select the command Structure of ring test in the menu Database or the buttons:
for the laboratory allocation
for the sample-measurand allocation
for the sample-measurand-laboratory allocation.

7.3.1 Laboratory allocation
In the window Laboratory allocation all laboratories that take part in the ring test can be selected.
In order to assign a laboratory to the ring test, double click into the column that is indicated
by a lighter background colour. If there are more ring tests in the database, they are shown
as well, so you can compare with previously organised ring tests. The current ring test is always shown in the first column and indicated by a lighter background. All remaining ring tests
follow in chronological order. By selecting rectangular sections of the column you can easily
choose several laboratories at once and either assign them to the ring test by clicking on the
button Include selected laboratories or delete them by clicking on Exclude selected laboratories. If you want to allocate all laboratories to the ring test, simply click on Include all laboratories.
Warning: Please note that deleting the allocations of laboratories will also delete the corresponding test results!

quo data GmbH

.

ProLab Plus – Manual Menu Database

33

Figure 16: Database – Structure of ring test – Laboratory allocation

By marking the checkbox Participating laboratories first, the laboratories that participate in
the corresponding ring test are listed first followed by the laboratories (if applicable) that do
not participate.
The table can be sorted by name, description or category by clicking in the respective column headers.
A list of all participating laboratories can be displayed by clicking on the Report button.
You can select the columns to be displayed, define the width of the columns and copy parts
or the whole table to the clipboard for further use, e.g. in Excel, with a right mouse button
click into the table.
7.3.2 Sample-measurand allocation
Select the menu Database – Structure of ring test – Sample-measurand allocation to specify
which samples have been or will be analysed with respect to which measurands.
To assign a measurand to a sample, check the corresponding cell by double-clicking. By selecting rectangular sections of the table you can easily choose several measurands at once
and either assign them to samples by clicking on the button Include selected combinations or
delete them by clicking on Exclude selected combinations. If you want to allocate all measurands to all samples, simply click on Include all combinations.
The current assignment table can be viewed as a report by clicking on the Report button.
You can select the columns to be displayed and define the width of the columns with a right

quo data GmbH

.

ProLab Plus – Manual Menu Database

34

mouse button click in the table. Furthermore, the whole table or sections of it can be copied
to the clipboard and pasted to e.g. Excel.
Warning: Please note that deleting the assignments of measurands will also delete the corresponding test results!

Figure 17: Database – Structure of ring test – Sample-measurand allocation

7.3.3 Sample-measurand-laboratory allocation
Select the menu Database – Structure of ring test – Sample-measurand-laboratory allocation
to assign samples and measurands to the laboratories of the ring test.

quo data GmbH

.

ProLab Plus – Manual Menu Database

35

Figure 18: Database – Structure of ring test – Sample-measurand-laboratory allocation (per sample)

You can now determine which samples are sent to which laboratories to investigate which
measurands. This can be accomplished by assigning laboratories one sample at a time, one
measurand at a time, for all measurands, or for all samples. The default setting when opening the window is per sample (i. e. one sample at a time).
If you choose one of the options per sample or per measurand, simply select the sample or
the measurand from the table on the right side and assign the laboratories to each measurand or sample by double-clicking in the respective cell. By selecting rectangular sections of
the table you can easily choose several samples or measurands at once and either assign
them to laboratories by clicking on the button Include selected combinations or delete them
by clicking on Exclude selected combinations. If you want to allocate all samples or all
measurands to all laboratories, simply click on Include all combinations.
It is also possible to assign laboratories to all measurands or all samples. If you choose
across all measurands, any specifications will refer to all measurands, i. e. they will apply to
all measurands that have been assigned to the particular sample under consideration.
If a laboratory does not analyse a certain sample, there will be no entry in the corresponding
field.
If only some of the measurands have been assigned to a laboratory (through the sample under consideration), the symbol % will be displayed in the field.
quo data GmbH

.

ProLab Plus – Manual Menu Database

36

If all measurands have been assigned to a laboratory (through the sample under consideration), an X will be displayed in the field.
You can create new assignments or delete existing ones. Any such changes will, however,
from now on affect all measurands that have been assigned to the sample in question.

Figure 19: Database – Structure of ring test – Sample-measurand-laboratory combinations (across all
samples)

By selecting rectangular sections of the table you can easily choose several samples at once
and either assign the combination of these samples and all measurands to laboratories by
clicking on the button Include selected combinations or delete them by clicking on Exclude
selected combinations. If you want to allocate all combinations of samples and measurands
to all laboratories, simply click on Include all combinations.
If you choose across all samples, any specifications will refer to all samples, i. e. they will
apply to all samples that have been assigned to the respective measurand. Again the symbols % and X are used to indicate that only some of the samples have been assigned to the
respective laboratory or that all samples have been assigned, respectively.
You can select the columns to be displayed and define the width of the columns with a right
mouse button click in the table. Moreover, the whole table or sections of it can be copied to
the clipboard and pasted to e.g. e.g. Excel.
The sample-measurand-laboratory combinations can be displayed as a report by clicking on
quo data GmbH

.

ProLab Plus – Manual Menu Database

37

the Report button.
Warning: Please note that deleting laboratory assignments will also delete the corresponding
test results!

7.4

Encoding

If the identity of the laboratory must not be revealed, the laboratory name must not appear in
the evaluation reports and has to be encoded in a suitable way. Apart from encoding the laboratories you can also encode the samples. This is useful if it is suspected that the laboratories may exchange their analytical results.
7.4.1 Encoding laboratories
All test results contained in the database are registered under an “address” consisting of the
four components ring test, sample, measurand and laboratory. To make sure that all data
can be correctly attributed at any time, this address must not change3, i. e. the particular laboratory always has to be indicated by LName. If, however, the identity of the laboratory
should not be revealed, LName must not appear in the evaluation reports and has to be encoded in a suitable way. The encoding can be done via Database – Encoding – Laboratories
or by clicking onto the button

in the main menu.

Any code can be entered in the Code field or laboratory codes can be generated automatically by clicking onto this button
. Existing codes are thereby deleted irrevocably. Please
note that the encoding only applies to the currently selected ring test.

3

This means that, for example, in order to execute an evaluation over various ring tests, a physical laboratory
should have the same value for LName in each ring test. If you change LName in the basic tables the allocation
will not be disturbed, as LName will be changed for all ring tests and all secondary windows.

quo data GmbH

.

ProLab Plus – Manual Menu Database

38

Figure 20: Database – Encoding – Laboratories

7.4.2 Encode test portions
Apart from encoding the laboratories you can also encode the test portions. This is useful if it
is suspected that the laboratories may exchange their analytical results. You can get to the
encoding via Database – Encoding – Test portions or by clicking the button
in the main
menu. Test portion codes are composed of up to 8 alphanumeric characters without special
characters and spaces.
Therefore the following notations are introduced: Sample denotes different kinds of samples
as water sample or soil sample. From each sample, test portions are produced which will be
sent to the laboratories or used for the homogeneity analysis.
Encoding can be performed
manually
automatically, without homogeneity test portions
automatically, taking into account the homogeneity test portions (see section 8.2)

Manual encoding
In order to encode the test portions an existing ring test structure is essential, i. e. data for
every sample-measurand-laboratory combination of the current ring test must exist. A code
can be assigned to each laboratory-sample combination. To encode the test portions unambiguously, any combination of test portion code and LName must exist only once. A violation
of this rule will prompt an Index Error, which can be adjusted by assigning a new code to the

quo data GmbH

.

ProLab Plus – Manual Menu Database

39

test portion. Please note that same samples in different laboratories may have same test portion codes.
Example:
SName LName Test portion code
Soil2
L01
Sample2
Soil2
L02
Sample2
Soil3
L02
Sample2

ok
ok, because Sample2 differs in LName
invalid, because Sample2 has already
been assigned to Soil2/L02

The test portion codes are entered via a matrix having the samples as columns and the laboratories as rows. In case of an invalid entry, an error message will be displayed upon leaving the edited field and the original value will be restored.

Automatic encoding
For convenience, the test portion codes can be generated automatically. First define the
number of test portions that will be generated per sample. If there are as many test portions
as participating laboratories, simply enter the number of laboratories. If there are more test
portions generated than participating laboratories ( e.g. if a homogeneity and/or stability test
is planned or if the remaining test portions will serve as reference material after the ring test),
simply enter the total number of test portions that are generated. By clicking on the button
, test portions of every sample are randomly assigned to each participating laboratory, and
a unique four-digit code is generated for every sample-laboratory combination. If you require
the test portion codes to be greater than a certain minimum value, enter this desired minimum value in the field First value for test portion code.
If the number of test portions per sample is less than the number of participating laboratories
(which should be the minimum number of test portions to generate), ProLab Plus generates
the codes on the basis of this minimum number.
If a homogeneity test is carried out, the respective test portions can be encoded by means of
the tab Homogeneity test portion encoding or later on under the menu item Homogeneity/Stability.
If originally no homogeneity test portions were included in the number of test portions per
sample, but homogeneity and/or stability test portions are nevertheless needed, it is recommended to enter the corrected total number of test portions per sample (i. e. the number of
test portions for the laboratories + the number of test portions for the homogeneity test + the
number of test portions for the stability test) and repeat the encoding of the test portions for
the laboratories in order to ensure a random ordering of all test portions.
In the rows Prefix und Suffix additional tokens can be entered, that will precede the code or
be attached to the end of the code, respectively.

quo data GmbH

.

ProLab Plus – Manual Menu Database

40

Figure 21: Database – Encoding – Test portions

Additionally there is the possibility to display a table with the test portion codes as a report
and to print labels with laboratory and sample specifications and optionally a barcode (where
ring test name, lab code and test portion code are coded). In order to do this, the label forms
in File – Settings – Reports – 75 Label / 77 Label with barcode must be adapted to the required label format.

Figure 22: Label for test portions with and without barcode

7.5

Participating labs and data yet to be submitted

In the menu Database – Participating labs and data yet to be submitted the following information is stored automatically or can be entered manually:
-

Date of sample arrival in laboratories
Date of measurements

quo data GmbH

.

ProLab Plus – Manual Menu Database

-

41

Date of the ring test organiser’s receipt of test results (reporting date)
Date of importing test results in ProLab
Persons involved
Comments.

Figure 23: Database – Participating labs and data yet to be submitted

All information can be viewed for the currently opened ring test or all ring tests that have
been made accessible (see section 5). Select Current ring test only or All unlocked ring tests.
If all that is required is an overview of the laboratories for which no test results have been
imported, check Only laboratories with data yet to be submitted.
A report with an overview can be generated listing either all laboratories or only those laboratories with results yet to have been received or imported.

7.6

Characteristics of measurands

After the ring test structure has been defined, the Characteristics of measurands dialogue
box allows further specification of the measurands analysed in the ring test. Reference values, reference standard deviations, measurement units and appertaining multiplication factors can be entered. Additionally two fields Rel. s.d. min and Rel. s.d. max for minimum or
maximum values of the relative standard deviations (in %) are available. In mode “L = limited” (see section 10.2.3) these values for the relative standard deviations will be used in the
assessment of the laboratories.
Using the button

all values – except for the reference value and reference s.d. – are as-

signed to the selected measurand in all samples of the ring test. Using the button
all
values – except for the reference value and reference s.d. – are assigned to all measurands
of the current sample. Using the button

all values – except for the reference value and

quo data GmbH

.

ProLab Plus – Manual Menu Database

42

reference s.d. – are assigned to all measurands in all samples of the ring test.

Figure 24: Database – Characteristics of measurands

If the database already contains units, they can simply be selected from the scroll-down
menu. The multiplicator, if included in the database, will automatically be selected by ProLab
Plus. New units must always be created via the window Characteristics of measurands.
Please click the button
next to the Unit field to open the Units window. In the table, the
new unit with its corresponding multiplicator can be entered. Please make sure to always
leave one “empty” unit, so the declaration of a unit can be suppressed if necessary.
The multiplication factor (multiplicator) is the factor, by which the test results have to be multiplied to obtain the concentration values. The multiplication factor for the measurement unit
µg/kg, for instance, is 10-9 = 1E-9, the multiplication factor for the measurement unit % is
10-2 = 1E-2. The entry of this multiplication factor is indispensable to guarantee the correct
calculation of the standard deviation according to Horwitz.

quo data GmbH

.

ProLab Plus – Manual Menu Database

43

Figure 25: Database – Characteristics of measurands – Units

7.7

Test results

To enter the test results, open the Test results dialogue box from the Database menu. You
can choose between entering the test results in a detailed form or in a table.
Depending on the ring test design that was selected under Database – Basic tables – Ring
tests the corresponding number of fields for entering test results is shown.
If no specific design was defined, the number of replicates can be manually determined in
the present window. The field Show test results makes it possible to define the number of
replicates (1 to 32) and, accordingly, the number of input fields for the test results.
If the ring test is of nested design the corresponding specific order of the test results must be
respected. For example, if for three different days two measurement values are available, the
design should be nested 3 x 2. There is a maximum of six test results per laboratory, the first
two belong to day 1, test results 3 and 4 to day 2 and test results 5 and 6 to day 3. The test
results need to be entered in this order.
Besides the fields for the test results, there are State fields as well. You can choose between
the following entries in these fields
NN
NB
GL
NA

=
=
=
=

not detectable,
not determinable,
greater than (Greater Limit),
not available

These can be used for the assessment of a laboratory. However, a State entry is not required, even when there are no test results available.

quo data GmbH

.

ProLab Plus – Manual Menu Database

44

Figure 26: Database – Test results – Detailed form

In the detailed form, the following shortcuts have been implemented, for the sake of convenience:
[Ctrl]+ [P]
[Ctrl]+ [M]
[Ctrl]+ [L]

previous sample
previous measurand
previous laboratory

[Alt] + [P]
[Alt] + [M]
[Alt] + [L]

next sample
next measurand
next laboratory

Apart from the test results, the limit of detection, the limit of determination, the upper limit of
the measuring range, the measurement uncertainty and the sample preparation and analytical method can be entered as well. In order to assign the sample preparation and/or analytical method for the measurements, they must have been defined beforehand. This is carried
out under Database – Definition of analytical methods and sample preparations (see section
7.9).
In the detailed form, additional information on the measurands ( e.g. Unit, multiplicator,…) as
entered under Characteristics of measurands (see section 7.6) are also shown. This information can be changed, but please be aware that these changes are specific to the measurand and will affect the test results of all laboratories.
Click
to get to the table view. The fields for the test results are named M 1, M 2, …, and
the fields for state entries are named S 1, S 2,… by default.
In order to adjust the layout to your needs, you can click the option Visible columns.

quo data GmbH

.

ProLab Plus – Manual Menu Database

45

Figure 27: Database – Test results – Table view

In the status bar at the bottom edge of the ProLab window, the description of the current laboratory, that of the sample and, if applicable, the test portion code are shown at all times.

7.8

Creation of corrected measurands

With ProLab Plus, it is possible to correct the measurement values and to generate a new
measurand (the old ones will not be overwritten) in the three following ways:

100
correction value
100
2. measurand
(100  correction value)
3. measurand  correction value

1. measurand 

The correction value may either be an already existing measurand or a constant that has to
be entered manually.

quo data GmbH

.

ProLab Plus – Manual Menu Database

46

Figure 28: Database – Creation of corrected measurands

Enter a name for the new, corrected measurand and click on the Start button to generate the
new measurand. A pop-up window will appear where you can enter a description of the new
measurand. After saving this description, a second pop-up window will appear where the
characteristics of the new measurand, such as unit, multiplicator, reference value, s.d., and
minimum/maximum rel. s.d. can be entered.

Figure 29: Pop-up windows after generating a new, corrected measurand

7.9

Analytical methods and sample preparations

Data on the sample preparation (SP) and the analytical method (AM) can be entered in the
database in case the influences of SP and AM are to be included in special statistical evaluations. Under Database – Definition of analytical methods and sample preparations all possible methods can be entered in two tables. An allocation to the actual measurements is carried out when entering the test results (see section 7.7).

quo data GmbH

.

ProLab Plus – Manual Menu Database

47

Both tables allow entry of a code (ID, only numbers are allowed), an abbreviation (Name,
used in the Summary results chart) and a full name (Description).
When defining new methods, click in the first row of the table and enter the corresponding
data. ProLab Plus will automatically provide missing entries, so it is sufficient to enter, for instance, only the description.
To generate a printable overview, click the Report button above the table; with a right mouse
click into the table, other available exportation means can be accessed.

Figure 30: Database – Definition of analytical methods and sample preparations – Analytical methods

quo data GmbH

.

ProLab Plus – Manual Menu Database

48

7.10 Duplicate ring test
Under Database – Duplicate ring test, a copy of the currently opened ring test can be generated. It is possible to choose between copying only the structure of the ring test or the whole
ring test including the test results. This feature is especially helpful when a ring test with the
same structure is carried out several times.
The name (8 digits) of the new ring test must be entered in the appropriate window. A more
detailed description can be specified in the Basic tables afterwards.

Figure 31: Database – Duplicate ring test

7.11 Export
ProLab Plus offers several export functions:
Export the whole ring test including the test results as a CSV file
Create laboratory files that can be filled in by the laboratories
Export the whole ring test as an LZH file
Contact laboratories by emails with attachments from ProLab Plus.
All functions are described in more detail in the following sections.
7.11.1 Test results as CSV file
Open the Database – Export – CSV file dialogue box to export test results and laboratory data. The export of data is recommended if the data is to be further processed in another program, such as EXCEL, or if a backup copy in ASCII format is to be created, that can later be
re-imported if necessary. The export is always carried out for the current ring test, unless the
box All ring tests is activated.

quo data GmbH

.

ProLab Plus – Manual Menu Database

49

Figure 32: Database – Export – CSV files

If only the selected data records should be exported, check Show selected data only. This
will reduce the number of lines in the exported file. (Of course, the desired fields need to be
selected.)
The following data are always exported: ring test name (RName), sample name
(SName = PName), measurand name (MName) and laboratory name (LName).
If All fields (columns) is checked, all information of the ring test(s) will be exported. If it is not
checked, fields such as laboratory mean (M), laboratory standard deviation (s.d.), Score, the
quantification limit (QL), detection limit (DL), the upper limit (GL), outlier values, additional
comments, sample preparation (SP), analytical method (AM), measurement unit, laboratory
and test portion codes, measurement uncertainty (MU) and recovery rate can be selected. If
the measurement results and corresponding state fields of the ring test should be exported
as well, click the button Measured values… to specify which of the results should be exported.

quo data GmbH

.

ProLab Plus – Manual Menu Database

50

Figure 33: Selection of measured values to be exported

After making your selection, click on the button Export to start the data export. A dialogue
box will appear, and you will be asked to enter the path and name of the export file.
The file generated by ProLab Plus can be imported into EXCEL in the CSV format. By default, the semicolon is used as column (field) separator so that the exported test result table
can easily be re-imported into ProLab Plus. For other applications it might be more appropriate to choose another field separator character from the list on the left of the export button.

7.11.2 Laboratory files
To avoid a laborious and therefore error-prone manual input of the laboratorial data, ProLab
Plus provides a function for automatic data transfer via CD or email (using appropriate security tools) between the participants and the organizer.
In order to prepare the data transfer, specific files can be created using the menu Database –
Export – Laboratory files (for RingDat). These are then filled in by the participating laboratories with the help of a special program called RINGDAT3. The data set for each laboratory
contains a FILENAME.LA2 file and a FILENAME.LAB file. The latter is an ASCII file with the
columns sample name (SName) / test portion code (SCode), Measurand name (MName)
and measurement unit. The columns containing the individual test results are added in the
RingDat3 program. FILENAME.LA2 is a configuration file that contains information on the
ring test, instructions and other parameters. The content of this file can be viewed in any text
editor; changes should, however, be avoided.
Besides the laboratory files, the participants also need to receive the following files, which
are located in the sub-directory RingDat3 in your current ProLab directory (the folder where
ProLab is installed):
RingDat3.EXE,
RingDat_English.HTML
translations\English.xml
As experience has showed that sending exe-files via email may cause problems, there is the

quo data GmbH

.

ProLab Plus – Manual Menu Database

51

possibility for the ring test participants to download all three files from the quo data homepage. Please contact us ([email protected]) to receive login data for the download.
In addition, ProLab Plus offers the possibility of creating individual laboratory files for each
laboratory and to save these on different storage devices ( e.g. CD, pen drive). To do this,
click on Copy labwise to storage device in the Create laboratory files (for RingDat) window. It
is now possible to select a laboratory and the drive (storage device) you would like to copy
the files to. Upon clicking copy or pressing F5, ProLab Plus saves the files FILENAME.LAB,
FILENAME.LA2, RingDat3.EXE, and RingDat_Englisch.HTML to the selected drive (see
7.11.2.1).
Furthermore, we can supply a customized online input tool for participants, thus circumventing the need to use RingDat for laboratory files.
The laboratory files are usually created lab-wise, though it is sometimes advantageous to
create one single file for all laboratories. If the check box per laboratory has been marked,
one set of data will be created for each laboratory. In this case FILENAME corresponds to
LName. If the check box is not marked, only one set of data with the RName as FILENAME
will be created. If the boxes per laboratory and Lab code are marked, additional files where
the FILENAME corresponds to the lab code are created. While the lab-wise option is easier
to fill out for the participants, the option for all laboratories simplifies the distribution of the
empty files to the laboratories and the following import.
If the check box Export selected combinations only is marked, only those files are exported
which correspond to the selected combinations of sample, measurand and laboratory (see
also section 9). If laboratories should specify the date of analysis, please mark the corresponding check box. The number of measurement values needs to be specified in order to
have sufficient fields available in the RingDat file. If it is required that a laboratory enter a
value in each field, please mark All data are needed. If the field Addresses is checked, there
will be an additional tab in RingDat, where every participant can see and edit his address information. If changes have been made, they will be highlighted during the import (see 7.12.2)
and can be incorporated into the Basic tables if desired.
When creating lab-wise files while the option Test portion code is enabled, a test portion
code has to exist for each laboratory and each sample. If the test portion code is missing,
the corresponding sample-measurand combination will not be displayed! When creating one
file for all laboratories only the names of the measurands and the measurement units will be
included in the file. The test portion code has to be added by the laboratory.

quo data GmbH

.

ProLab Plus – Manual Menu Database

52

Figure 34: Database – Export – Laboratory files (for RingDat)

To transfer information about the analytical methods and sample preparations, please activate the check box Analytical methods and sample preparations. In this case two additional
tabs will be displayed. On the tab Select AM and SP the analytical and sample preparation
methods to be made available in the program RINGDAT3 can be selected. If the list of available analytical methods is long and unwieldy, but not all analytical methods are meaningful
for any particular measurand, there is the possibility to deselect meaningless analytical
methods on the tab Allocation of AM to measurands.

quo data GmbH

.

ProLab Plus – Manual Menu Database

53

Figure 35: Database – Export – Laboratory files (for RingDat): Allocation of AM to measureands

A text box is available for entering information for the laboratories (General information for
laboratories). RINGDAT3 will display the contents of this text box at the bottom of the Measured values tab. Furthermore, in RingDat there is the possibility to print a protocol of the test
results. It is possible to enter information for the laboratories to be displayed on these protocols as well. This information will be shown on the header on the left and/or on the right.
Make sure, the box Include general information in protocol is checked.
Before exporting, the order in which the combinations of samples and measurands are sorted in RingDat must be determined: according to the samples first and then the measurands
or vice versa.
Finally, via the Export button and the dialogue box which appears, all lab files can be saved
in the folder you select.. The *.lab files can now be opened with RingDat3.exe and measurement values can be entered.

quo data GmbH

.

ProLab Plus – Manual Menu Database

54

Figure 36: Lab file for laboratory L005 opened with RingDat3

Copy labwise to storage device
In this tab you can select a laboratory from the list on the left, a target drive (e.g. a pen drive)
from the drop-down menu on the right and click on the Copy button.
When the pen drive is removed from the computer, the input field Target drive will be empty.
If Windows uses the same target drive letter again, as soon as a new pen drive is plugged in,
the list in ProLab will show it again. If there are more than one pen drives plugged in, Windows will choose a different drive letter for each one.
ProLab will store the creation date of the files copied onto the pen drive for each laboratory.
That information can be seen via Database – Participating laboratoriess and data yet to be
submitted as well.
On the sidebar there is a list of files to be copied onto the storage device. You can add your
own files by clicking onto the list using the right mouse button.

7.11.3 Ring test
Besides the export of certain data (especially structure of ring test and test results) into an
ASCII file, there also exists the possibility to save a ring test as a complete database. In order to do this, copies are generated of all important tables in the current database. These tables contain only data of the selected ring test. They are denoted by the prefix ‘~’ and saved
in an LZH table. Importing a ring test is carried out via Database – Import – Ring test (LZH
compressed). Thereby it is possible to archive data of a ring test, delete the ring test from the
current database and import it at a later point in time.
Please take note of the advice on long-term storage.

quo data GmbH

.

ProLab Plus – Manual Menu Database

55

7.11.4 Email merges
In ProLab Plus there is the possibility to send email merges to the participants of the ring
test. The email merge module can be found under Database – Export – Email merge.
This is applicable for sending e.g.
General information emails, such as announcements of meetings, delays and other administrative information
Deadline reminder emails
Single laboratory files including RingDat
Downloads
Analysis results and certificates
An important prerequisite for using the email merge module in ProLab Plus is the entry of
email addresses in the basic tables of the database. More than one address can be specified
by using spaces or semicolons as delimiters. Additional data fields, such as laboratory name,
laboratory description, and the title of the contact person, are relevant as well.

Figure 37: Database – Basic tables – Labs as Table: Enter relevant information

The first tab of the email merge window is called Compose email and this is where the email
text can be entered, and the attachments specified. This is also where it is possible to determine which laboratories should receive the email.

quo data GmbH

.

ProLab Plus – Manual Menu Database

56

Figure 38: Database – Export – Email merge: Compose email

quo data GmbH

.

ProLab Plus – Manual Menu Database

57

Template files
There is the possibility to import and use templates for a email merge. To do so, specify the
path of the file (.txt or .rtf) after clicking Import text file. In this text file, the only requirement is
that the first line contains the subject of the email. The main body of the email then starts in
the second line, typically beginning with the addressee’s title and name before proceeding
with the content of the email.

Figure 39: Import of individual templates from a text file

After selecting a template (see example in Figure 39), the text file is transferred to the editable text window. It is also possible to insert macros, that is to say, variables that stand for information that changes from laboratory to laboratory and that is then entered in the email automatically for each participant; for example, the addressee’s name and title, or the laboratory’s code, In order to insert a macro, mark the section in the email text where the it should
appear and select the macro from the box Macro selection.

quo data GmbH

.

ProLab Plus – Manual Menu Database

58

Attachments
Attachments can be included as static attachments, which are the same for all laboratories,
or as dynamic attachments, that is to say, attachments with laboratory-specific information
that are only sent to the corresponding laboratories.
In order to include a file as a static attachment, click on the button Add file ( ) below the Attachments field and select the desired file. This static file is then sent to all laboratories, e.g.
the final report which all laboratories should receive. The selected path appears under Attachments.
A precondition for the attachment of dynamic files is, of course, that these laboratory files,
which differ only in their laboratory name or description, have to be available. These different,
laboratory-specific files will also have laboratory-specific names, or laboratory-specific paths.
To start, select one such file via the Add file button ( ). The corresponding path is now
shown in the Attachments field. If you click on it, it will appear in the Edit field (orange) below
the Attachment field. In the next step, mark the section in the file path where the macro
should be inserted. Finally, select the desired macro from the box Macro selection.
Single attachments or all of them can be deleted via the buttons Remove selected file and
Clear attachment.

Figure 40: Attaching a dynamic file to the email

Of course, the laboratories which should receive the email need to be selected as well. This

quo data GmbH

.

ProLab Plus – Manual Menu Database

59

is done in the right part of the tab Compose email. In this box, all laboratories that were supposed to participate in the current ring test are shown, i. e. all laboratories that were selected
for participation. However, it might be that not all of these laboratories have submitted data.
Thus, it is possible that not all of these laboratories should receive a particular email with attachments. For example, a report with individual results will not be available if a laboratory
has not submitted any data.
Laboratories to which emails should be sent can be selected manually by marking the respective boxes. Alternatively, all laboratories can be selected in one step (Select all labs).
It is also possible to base the selection of laboratories on the existence of PDF files. In order
to do this, there must be a folder with PDF files whose names refer to laboratory names. For
example, one PDF file could be called “Labmean (Merkmal) L001.pdf” (see Figure 41). Using
the button Selection by serial PDF files and choosing the relevant folder will lead to a selection of only those laboratories referred to in a PDF file name in the chosen folder.

Figure 41: Selecting laboratories by PDF files

On the tab Preview, the email merges are displayed. All inserted macros are replaced by the
corresponding data. The preview window shows one email after another with the individual
attachments.

quo data GmbH

.

ProLab Plus – Manual Menu Database

60

Figure 42: Preview of email merge

The details regarding the email server are displayed on the last tab Settings. Appropriate settings should be carried out by the administrator.
Before the emails are sent, an automatic data consistency check (e.g. the existence of all
files to be sent as attachments) is carried out. It is also possible to carry out such a consistency check manually in the Compose email tab by using the button with the magnifier.
ProLab Plus will tell you if the check was successful or not.
To send the emails, there are two possibilities: either (1) all emails are first saved in the outbox of the email program (Outlook, Thunderbird etc.) or (2) all emails are directly sent via the
mail server of the provider. The first option has the advantage that all mails can be checked
again with the email program and are listed in the Sent folder after they have been sent.

quo data GmbH

.

ProLab Plus – Manual Menu Database

61

Figure 43: Database – Export – Email merge: Settings

7.12 Import
ProLab Plus allows the import of
-

The structure of the ring test with or without ring test results
Laboratory addresses
Measurands & samples with all characteristics
Analytical methods & sample preparations
Ring tests (LZH compressed)

Click in the menu item Database – Import and select the type of data to be imported. In the
following, the import of measurement data will be described in detail. The import procedure
of addresses or information on the measurands and samples is similar.
7.12.1 Import of structure with or without test results
Almost all data that can be entered manually into the database may also be imported from
EXCEL or CSV files.
The general structure of the import files is always the same: each column has to be assigned
to a field in the database. The import procedure comprises the following steps:
1. select type of data to be imported,
2. open the import file or paste the data into the import window,
3. assign columns in the import file to database fields,
4. start transfer of the file into the database.
Click on

to open the import file. In the opening dialogue box select the correct directory,

quo data GmbH

.

ProLab Plus – Manual Menu Database

62

file type (*.db, *.csv, *.lab, *.txt, *.xls, *.xlsx) and name of the import file. Excel files need to
be saved as EXCEL 97 or above. The file to be imported must not be opened or used by another program. Alternatively, input via copy & paste is possible by either clicking into the
empty table and pressing the key combination Ctrl + V, or by right-clicking and selecting the
option Paste from the opening menu.
Subsequently the individual columns are assigned to the database fields. In order to select
these fields, please click on the arrow in the top row of the relevant column. A list will appear
where you can select the database fields. The fields M 1, M 2, …, M 32 correspond to the
test results 1, 2, …, 32.
If the ring test is of nested design a certain order of the test results to be entered needs to be
respected. For example if for three different days two measurement values are available, the
design should be nested 3 x 2. Thus, in total there are up to six test results, the first two belong to day 1, test results 3 and 4 to day 2 and test results 5 and 6 to day 3. The test results
must be entered in that order.
Columns that are already assigned to database fields are highlighted in blue.

Figure 44: Database – Import – Structure/ring test results

Before the data is transferred into the database, all rows that do not contain valid measurements (e.g. the headline) should be deleted. This can be done by clicking with the right
mouse button on the cell, column or row and selecting Delete.
If the first row contains the headers that should not be imported, simply check the box Ignore
first line on the right side of the window.
The correct decimal delimiter in accordance with your data has to be chosen on the right side
of the window before starting the transfer.

quo data GmbH

.

ProLab Plus – Manual Menu Database

63

The imported test results are always assigned to the currently selected ring test. If test results should be imported into a new ring test, please create a new ring test in the menu Database – Basic tables and select it for editing. Afterwards test results may be imported as described above.
If the import file only contains one laboratory, no column for lab names is required. In this
case, mark the check box Laboratory on the right side of the window and enter the name of
the laboratory. Proceed analogously if the import file only contains pertaining to one sample
or measurand. The name of the laboratory or sample is entered into the corresponding field.
This name is then assigned to all datasets of the file in question.
It is further to be noted that there are two columns each for sample, measurand and laboratory in the basic tables. Those columns refer to name and description. As the name must consist of no more than eight characters (alphanumerical or underscore), the import into ProLab
Plus works as follows:
1. A name for each dataset is created by reducing the already existing text to eight
characters, after removal of all characters that are not allowed.
2. If there is no already existing sample (or measurand or laboratory) with this name, it is
applied. In this case, and in this case only, the entire original text will be used as description.
3. If the name of a sample (or measurand or laboratory), that has already been imported
during the current process, is the same as the newly generated one, ProLab Plus determines whether it corresponds to the same text. For example, since “A+” and “A-“
are assigned the same new name “A”, they would be treated as the same sample (or
measurand or laboratory). In case of such a collision the user will be asked if both
samples can be assumed to be identical.
It is recommended to use the semicolon as column (field) separator. Depending on the Windows setting, the use of comma or point is recommended as decimal separator.
After checking the plausibility of the table with respect to correct column assignment, start
the transfer of data by clicking on the button Transfer data. Now all columns that contain values attached to a database field (highlighted in blue) are entered successively in the database. Since an automatic comparison of the different tables is performed at the same time,
the import of the file may take a few minutes. The changes within the database and error
messages will be listed in the log window and the log file (see chapter 6.3).
Please note: Be aware that wrong names or column assignments for the laboratories, samples, measurands or ring tests may have serious consequences for the database, and that
restoring the original database or repairing the damage may take great effort. It is therefore
important to thoroughly check the names and the assignment of the columns before starting
the import. Moreover all tables of the database should be backed up first. For example, if a
laboratory with name “001” has been defined in the Basic tables but, in the process of importing data, the name of the lab is changed to “1”, ProLab Plus will interpret this as a new laboratory and will not assign the values to laboratory “001”. Thus, make sure the right format is
defined in the Excel table (showing leading zeros etc…)..
You can also import data if the corresponding laboratories, ring tests, samples and measurands have not yet been defined. In this case ProLab Plus will ask you if a cross-check of the
quo data GmbH

.

ProLab Plus – Manual Menu Database

64

tables should be performed. Click the All button to cancel the enquiry, which will otherwise
reappear at every new entry.
Remember that descriptions cannot be entered in the Import menu, but only manually in the
Basic Tables menu.
If there is a space character in between two separators, ProLab Plus will interpret this as a
column that contains a NULL value.
Test results that contain invalid characters will result in a warning.
Test results with the character '<' will be interpreted as results below the limit of quantification. In this case the test result will be set to NULL and obtain the state NB. The actual value
will be entered as the limit of quantification.
Test results with the character '+' will be interpreted as results above the “greater limit”. In
this case the test result will be set to NULL and obtain the state GL. The actual value will be
entered as “greater limit”.
7.12.2 Addresses, Measurands & samples, Analytical methods/Sample preparations
The import of laboratory addresses, samples and measurands, and of analytical methods
and sample preparations is analogous to the procedure described in section 7.12.1.
7.12.3 Ring test
The basis for importing a ring test is an LZH-file that has been created earlier by exporting a
ring test via Database – Export – Ring test (LZH compressed).

Please note that already existing data with identical names (whether ring tests, samples, measurands or laboratories) will be overwritten after confirmation is given. The
confirmation prompt will appear even if there are no values in the actual database but
the fields are already defined.

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

65

8 Test on homogeneity and stability
8.1

Import test results

This window allows the import of homogeneity and/or stability test results from an Excel file.
The import is carried out as described in section 7.12.1.
If you want to enter the homogeneity test results manually you need first to encode the test
portions (see section 8.2).
If only a test on homogeneity should be carried out, the Excel table has to contain columns
for the sample name, the measurand, the test portion code and the test results. If a stability
test should be carried out as well, an extra column for the date is necessary in order to distinguish homogeneity and stability measurements.
Sample
tuna
...

Measurand
As


Test portion code
1
...

M1
2.83
...

M2
3.14
...

At first, the Excel table needs to be opened in ProLab Plus. Accordingly, it must not be already opened with another program. The data may also be entered via copy & paste. In the
next step, the columns need to be assigned and further import specifications need to be carried out before the import can be started.

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

66

Figure 45: Homogeneity/Stability – Import test results

8.2

Encoding test portions

The window Test portion encoding offers the possibility to encode all test portions and select
homogeneity test portions randomly.
For the sake of clarity, the following terminology is introduced: sample denotes different kinds
of sample, such as water sample or soil sample. From each sample, test portions are produced which are sent to the laboratories, used for the homogeneity analysis or kept behind
as retain test portions or for stability analysis.
Furthermore, it has to be determined if the test portions for the laboratories (i. e. for the ring
test) and for the homogeneity analysis are to be encoded uniformly, or if the test portions for
the homogeneity analysis should be encoded separately.

8.2.1 Uniform encoding of all test portions (for laboratories and for homogeneity
analysis) and random selection of homogeneity test portions
Encoding of all test portions is carried out in two main steps. First codes are generated for all
test portions and then some of these test portions are assigned randomly to the laboratories
and others to the homogeneity analysis. These encoding and selection steps are not performed in the menu Homogeneity/Stability but in the menu Database – Encoding – Test portions, because it is necessary to include the homogeneity test portions in the encoding from

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

67

the beginning or else a random selection of these test portions cannot be ensured. If the homogeneity test portions have not been considered from the beginning, the whole encoding –
including that for the laboratory test portions – should be repeated (see also section 7.4.2).
If there are several kinds of sample available, there is the possibility to assign the codes in
such a way that it is impossible to identify the kind of sample by means of the codes. If for
example 100 test portions are produced for each of two samples, all 2 x 100 codes can be
assigned to the test portions randomly over both samples instead of assigning codes 1 - 100
to the first sample and codes 101 - 200 to the second sample.
Via the menu Database – Encoding – Test portions, codes are generated at first for all produced test portions, for the laboratories as well as for the homogeneity analysis – the homogeneity test portions are selected later on. The codes may be generated manually or automatically.
The individual steps are:
1. Enter the total number of test portions produced per sample, e.g. 95 test portions of
which 80 (= number of laboratories) are shipped to the laboratories, 10 are used for the
test on homogeneity and the remaining 5 test portions serve as retain test portions
2. Generate codes for test portions for the laboratories on the tab Test portion encoding:
3. After encoding the test portions for the laboratories, the test portions for the homogeneity test have to be encoded. To do so, select the tab Homogeneity test portion encoding
or the menu item Homogeneity/Stability – Test portion encoding.
4. Enter the number of test portions for which the homogeneity analysis will be carried out,
e.g. 10 of the 95 test portions of each sample;
5. Generate the codes for the test portions of the homogeneity analysis via the button

Figure 46: Database – Encoding – Test portions: Steps 1 and 2 for encoding test portions

quo data GmbH

;

68

.

ProLab Plus – Manual Test on homogeneity and stability

After the codes have been generated, labelling of the test portions can be carried out. The
labels should be printed beforehand, as should the overview of the assignment of labels to
test portions. The labels can be edited individually before printing.
For each sample, the assignment overview as well as the labels are sorted in ascending order according to their code. Then, for the sample in question, labels can now be affixed successively to randomly selected test portions. The labelled test portions can then be sorted
according to whether they are to be shipped to the laboratories or are meant for the homogeneity analysis.

Figure 47: Assignment overview

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

69

8.2.2 Random selection of test portions for homogeneity test without encoding test
portions for laboratories
ProLab Plus allows the random selection of homogeneity test portions from all produced test
portions without having to encode all of them. This is done under Homogeneity/Stability –
Test portion encoding. First, the total number of test portions (per sample) needs to be specified, as well as the number of homogeneity test portions. In the case that more than one
sample are being tested, the codes will be spread randomly over all samples.
The proceeding is as follows:
1. Enter the total number of test portions (per sample), e.g. 100
2. Enter number of test portions per sample used for the homogeneity analysis, e.g. 10
3. Generate the codes for the random selection of the test portions for the homogeneity
analysis via the button

.

quo data GmbH

70

.

ProLab Plus – Manual Test on homogeneity and stability

Figure 48: Homogeneity/Stability – Encoding test portions: Steps 1, and 2 for encoding test portions

Figure 49: Homogeneity/Stability – Test portion encoding: Step 3 for encoding test portions

The test portions for the homogeneity test can now be selected according to the randomly
generated codes. Ideally, all produced test portions already have a unique number ( e.g. order of production) so that all that remains to be done is to select the test portions with the
corresponding numbers. If there is no such numbering, the selection may be made simply by
counting.
8.2.3 Another example for encoding test portions
Ring test organiser’s situation
100 laboratories will participate in the ring test. For these participants there are 115 test portions containing 230 g of material and another 100 test portions containing 700 g of material,

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

71

which can be used as reference material at the end of the test. So in total there are 215 test
portions. For the test on homogeneity 20 of those 215 portions should be selected. For the
first group (with test portions containing 230 g material) a maximum of 15 test portions may
be randomly selected, otherwise there would not be enough test portions for all participants
left.
Solution using ProLab Plus
In the first step the number of test portions in the window Database – Encoding – Test portions – Test portion encoding will be determined as 115. When clicking on the button with the
little key, the 100 test portions for the laboratories will get a randomly selected code in the
range from 1 to 115. After that – using the tab Homogeneity test portion encoding – the total
number of test portions will be raised to 215. In the next encoding step the test portions for
homogeneity tests will be randomly selected out of the remaining 15 portions containing
230 g and the remaining 100 portions containing 700 g.

8.3

Test results

Test results for the homogeneity and/or stability test may be entered manually in the window
Homogeneity/Stability – Test results. At the top of the window you can choose the sample
and measurand. You can enter up to 32 test results for each test portion. If it is necessary to
distinguish between homogeneity and stability results, a date must be entered.

Figure 50: Homogeneity/Stability – Test results

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

72

8.4

Computation

8.4.1 Assignment of test portions
If data for both the test on homogeneity and the test on stability have been imported, it is
necessary to determine which data belong to the homogeneity analysis and which to the stability analysis. This is accomplished by specifying date ranges for both kinds of samples.
If no stability analysis has been carried out, no date range needs to be defined in order to
carry out the analysis of the homogeneity test results. In this case, simply all available data
are used in the statistical analysis. For the stability test, however, a date range always needs
to be defined. In that case it is absolutely necessary to include dates in the database.

Figure 51: Specifying date ranges for analysis of homogeneity samples and stability samples (in this exst
ample, the homogeneity analysis has been carried before 31 July 2010, whereas the stability samples
th
have been analysed after 30 September 2010)

8.4.2 Homogeneity
For the test on homogeneity, the following statistical methods are available:
F test:
Check for significant heterogeneity
ISO 13528:
Check for sufficient homogeneity
Harmonized Protocol: Test on significant heterogeneity4
4

M. Thompson, S. L. R. Ellison and R. Wood: Pure Appl. Chem., Vol. 78, No. 1, pp. 145–196, 2006.
doi:10.1351/pac200678010145

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

73

The necessary parameters for the test on homogeneity are the analytical precision
s(analytical) (standard deviation within the test portions) and the heterogeneity standard deviation s(samples) (standard deviation between the test portions).
All standard deviations can be displayed and used in the calculations as absolute or as relative quantities. The selection is done in the upper left part of the window.
The F test is used to determine whether the heterogeneity standard deviation s(samples) deviates significantly from zero to a significance level of 5 %, i. e. if it can be assumed that the
difference between the mean values are caused by measurement deviations only. If the latter
is the case, the assumption of sample homogeneity is accepted. If not, it can be concluded
that the sample is not suitable for a ring test.
For the method according to ISO 13528 and to the Harmonized Protocol, a target standard
deviation s(target) is necessary.
In ISO 13528, the heterogeneity standard deviation is compared to the target standard deviation. The samples can be considered sufficiently homogenous if the heterogeneity standard
deviation is not greater than 0.3 x target standard deviation. The motivation of the factor 0.3
is the fact that in the case of agreement between the target standard deviation and the reproducibility standard deviation (here, the heterogeneity standard deviation), the latter increases due to the sample heterogeneity by 10 % at most.
According to the Harmonized Protocol (2006), the hypothesis that the sample is sufficiently
homogenous is checked. In a first step, the analytical precision is analysed. This must be
smaller than 0.5 x target standard deviation. An insufficient analytical precision may conceal
significant sample heterogeneity. In a second step, the heterogeneity standard deviation is
analysed. If the latter exceeds the allowable quantity of 0.3 x target standard deviation to a
significance level of 5 %, the hypothesis is rejected, and the sample is deemed unsuitable.
The Harmonized Protocol does not make heavy demands regarding the sample homogeneity. Even if the heterogeneity standard deviation is significantly larger than zero and lies within
the range of 50 % of the target standard deviation, the samples might be accepted.
In ProLab Plus, at first the mean value, the analytical precision and the heterogeneity standard deviation are displayed in a table for each sample-measurand combination. The result of
the F test is shown as well: “OK” means that there is no significant heterogeneity; “Not OK”
means that there is significant heterogeneity.
For the test according to ISO 13528 and the Harmonized Protocol, the target standard deviation needs to be specified at first. This has to be carried out for each sample-measurand
combination. The following modes are available:
Horwitz: calculation of the target standard deviation according to Horwitz
Corrected Horwitz: if the relative target standard deviation according to Horwitz is greater than 22 %, it is truncated to 22 %
Manual: fixed quantity for target standard deviation, e.g. 10 %
Target standard deviation from ring test: applicable after analysing a ring test, i. e. after
the data of the ring test have been analysed statistically
For the calculation of the Horwitz standard deviation, a so-called multiplicator corresponding
to the measurement unit is necessary. Therefore, before calculating the Horwitz standard

quo data GmbH

74

.

ProLab Plus – Manual Test on homogeneity and stability

deviation, a unit and the respective multiplicator need to be specified for each measurand.
This is done under Database – Characteristics of measurands (see also section 7.6).
The Horwitz standard deviation as well as the corrected Horwitz standard deviation may be
further adjusted by the Horwitz Ratio (HORRAT). The HORRAT value is the ratio between
the reproducibility standard deviation of a ring test and the Horwitz standard deviation. A
HORRAT value smaller than 1 shows that the reproducibility standard deviation of the ring
test is smaller than the respective Horwitz standard deviation. HORRAT-corrected standard
deviations may be calculated by multiplying the Horwitz standard deviation by the HORRAT
value. This HORRAT-corrected Horwitz standard deviation may be used as an estimation for
the reproducibility standard deviation expected in the current ring test. If the current reproducibility standard deviation is larger than the HORRAT-corrected Horwitz standard deviation
computed on the basis of data from a former ring test, it may be an indication that specific
analytical difficulties have appeared which may have been caused by a different sample matrix.
The specification of the mode of the target standard deviation is carried out via the scrolldown menu in the column Mode s(target). If a HORRAT value is available, it can be entered
manually in the column HORRAT. As soon as a value is entered that differs from 1, the
HORRAT-corrected target standard deviation is calculated automatically. In the mode Manual the target standard deviation needs to be entered manually in the corresponding column.
Example:
The procedure is explained by the example in Figure 52:
For the measurand Arsenic the mode Horwitz has been selected and a HORRAT value of
0.75 has been specified. The relative target standard deviation is 0.75 x 13.69 % = 10.27 %.
For the measurand Copper the mode Manual has been selected and a relative target standard deviation of 20 % has been specified manually.
For the measurand Lead the mode Horwitz has been selected and the relative target standard deviation equals 17.78 %.
After the specification of the target standard deviation, the results of the test on homogeneity
according to ISO 13528 and the Harmonized Protocol are displayed.
ISO 13528
o OK = sufficient homogeneity
o Not OK = insufficient homogeneity
Harmonized Protocol
o OK = no significant heterogeneity
o Not OK = significant heterogeneity

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

Figure 52: Homogeneity/Stability – Computation – Homogeneity (table)

quo data GmbH

75

.

ProLab Plus – Manual Test on homogeneity and stability

76

For each sample-measurand combination a report and a figure is shown above the table.
The relevant sample-measurand combination must first be selected from the scroll down
menu in the middle part of the window.
The figure shows the mean values and the standard deviations within the test portions of one
sample-measurand combination.

Figure 53: Homogeneity/Stability – Computation – Homogeneity (chart)

For the report, two alternatives are available: a simple report according to the criteria of
VDLUFA5 as well as a report with a detailed statistical background. Both reports summarize
the results of the homogeneity analysis; however, the report with the statistical background
offers more extensive information.
The figure and the selected alternative can be displayed in a report via the Report button. If
the option labelled Table is selected, a table with the statistical parameters for all samplemeasurand-combinations is included in the report, whereas selecting the option labelled
Chart leads to a display of the report for the currently selected sample-measurandcombination.
8.4.3 Stability
ProLab Plus also offers the possibility to analyse measurements on the stability of the samples. It is assumed that, before the performance of the ring test, test portions were randomly
selected for homogeneity testing and that the mean value of these test portions was determined. This mean value is then compared to the mean value of retained stability test portions
which are analysed after a certain time span. Thus, for the stability test, measurement values
from a homogeneity test need to be available. If no homogeneity test has been carried out,
but e.g. a reference value is available with which the stability test portions should be compared, then it is this reference value which must be imported as homogeneity test result. Furthermore, it is absolutely necessary to give a date to every measurement that is performed in
the context of a stability analysis, since it is the presence or absence of a date which makes
it possible to distinguish stability from homogeneity measurements.
There are two different tests on stability available in ProLab Plus: the one based on ISO
13528, and that based on the Harmonized Protocol. For the method according to ISO 13528
5

Verband Deutscher Landwirtschaftlicher Untersuchungs- und Forschungsanstalten e.V. (Federation of Laboratories for Agricultural Research)

quo data GmbH

.

ProLab Plus – Manual Test on homogeneity and stability

77

a target standard deviation s(target) is necessary. Samples can be considered stable if the
mean value of the stability test measurements differs less than 0.3 x s(target) from the mean
value of the homogeneity test measurements. The Harmonized Protocol (t test) is also based
on a comparison of the mean values of the homogeneity and the stability test measurements.
The samples may be considered stable if there is no statistically significant (5 % significance
level) difference between the mean values.
If there are significant differences between the mean values, sample preparation and storage
conditions should be checked in order to identify reasons for the instability.
For the test according to ISO 13528, the target standard deviation must first be specified.
This has to be carried out for each sample-measurand combination. The following modes are
available:
Horwitz: calculation of the target standard deviation according to Horwitz;
Corrected Horwitz: if the relative target standard deviation according to Horwitz is greater than 22 %, it is truncated to 22 %;
Manual: fixed quantity for target standard deviation, e.g. 10 %;
Target standard deviation from ring test: applicable after analysing a ring test, i. e. after
the data of the ring test have been analysed statistically.
For the calculation of the Horwitz standard deviation, use is made of a so-called multiplicator
as appropriate to the particular measurement unit. Accordingly, before calculating the Horwitz standard deviation, a unit needs to be specified for each measurand. This is done under
Database – Characteristics of measurands (see also section 7.6).
In the last two columns of the table the results of the stability tests are shown:
OK: adequately stable
Not OK: not adequately stable

Figure 54: Homogeneity/Stability – Computation – Stability (table)

The chart below the figure shows the means and s.d. of the homogeneity test portions (denoted with “H”) and the stability test portions (denoted with “S”). The lines represent the
mean values of all homogeneity and all stability test portions. Selection of the sample and
measurand takes place in the boxes above the chart.
Beneath the chart, a report with a comprehensive presentation of the results of the stability
test is displayed (for the selected sample and measurand).

quo data GmbH

78

.

ProLab Plus – Manual Test on homogeneity and stability

Figure 55: Homogeneity/Stability – Computation – Stability (chart)

quo data GmbH

.

ProLab Plus – Manual Data selection

79

9 Data selection
9.1

Overview

The selection tool in ProLab Plus is useful for quick selection of subgroups of samples,
measurands or laboratories. Calculations and assessments may then be carried out separately for different subgroups. Charts could be much easier to understand if divided into different charts for different subgroups.
The specification of categories for laboratories, samples and measurands is very useful
when carrying out the selection. For example, when calculations are to be carried out separately
for EU and Non-EU laboratories
for organic or inorganic measurands or
for different analytical methods or sample preparations
The Selection menu provides two selection options
Sample / Measurand selection
Sample / Measurand / Laboratory selection
You can choose either option to select the data set that is to be considered for evaluation,
assessment and presentation. In each of these steps, the data set can be selected separately. For instance, to determine the assigned value and the target standard deviation, the evaluation may be based on selected reference laboratories, while all laboratories may be included in the assessment and the presentation of the results.
Up to four selection combinations can be saved and reloaded. Please note that both the
sample-measurand and the sample-measurand-laboratory selections are always saved at
the time.

9.2

Sample-measurand selection

The selection of ring test, sample and measurand can be found via Selection – Sample /
Measurand.
When selecting the data set, the hierarchical structure of the data has to be taken into account:
Ring test
Sample
Measurand
Laboratory
Thus before selecting a specific sample, the ring test has to be chosen first. Similarly, before
selecting a measurand, both the corresponding ring test and sample have to be selected
first. This makes it possible to select the measurands in a sample-specific way. You can display the sample-measurand combination for the currently opened ring test or for all ring tests
in the database that have been made accessible before. (For the latter option, mark the
checkbox All ring tests in the upper part of the window.) The first row of the table shows the
name of the ring test.
quo data GmbH

.

ProLab Plus – Manual Data selection

80

Figure 56: Selection –Sample-measurand (currently opened ring test only)

The following symbols are used in the overview of all activated sample-measurand combinations:
The corresponding sample-measurand combination is activated.
The corresponding combination is deactivated.
Grey

This sample-measurand combination does not exist, i. e. the measurand has
not been assigned to the sample.

By double-clicking on a desired combination, it can be activated or deactivated. You can also
select several fields and activate or deactivate them at once by using the following buttons in
the upper bar of the window:
Activate selection
Activate all
Deactivate selection

quo data GmbH

.

ProLab Plus – Manual Data selection

81

Deactivate all

The current assignment table can be viewed as a report by clicking the Report button. When
that button is clicked and the “Shift” key  is simultaneously pressed, the table may be saved
as an EXCEL file.
If the same selection is to be used in various ring tests, it is advisable to assign appropriate
categories in the Database - Basic Tables - Measurands menu. By selecting such a category, all corresponding measurands will be activated simultaneously. If you choose two categories, then they function as criteria which have to be fulfilled simultaneously for a measurand
to be activated. If you require only one criterion, simply leave the field for the second category empty. By clicking the button
all measurands that fulfil the selection will be activated,
clicking the button deactivates all combinations that fulfil the selection. For the selection of
samples via categories, the procedure is analogous.

Figure 57: Selection – Sample-measurand: selecting all heavy metal samples

9.3

Sample-measurand-laboratory selection

In order to select laboratories and measurands the menu Selection – Sample-measurandlaboratory can be chosen.
quo data GmbH

.

ProLab Plus – Manual Data selection

82

Please note that any laboratory selection will directly depend on the selected ring tests, samples and measurands. It is only possible to include in the selection test results for which the
corresponding sample and measurand have previously been selected in the Selection –
Samples -measurands window. After selecting the desired ring test and sample/measurand,
an overview of all activated laboratory/sample/measurand combinations will be displayed in
the table. Laboratories can be selected in terms of samples or measurands. If selection per
samples is chosen, all samples on the left side of the window must be selected successively.
Selection per measurands works analogously.

Figure 58: Selection –Sample-measurand-laboratories: selection for sample 01, sample preparation SP2,
analytical method AM6 and the category 1 are specified for a selection of the laboratories

The following symbols are used in the overview of all activated sample-measurand combinations:
The corresponding laboratory/sample/measurand combination is activated.
The corresponding laboratory/sample/measurand combination is deactivated.
Grey

This sample/measurand combination does not exist, i. e. the laboratory has not
been assigned to the sample/measurand.

By double-clicking on a desired combination, it can be activated or deactivated. You can also
select several fields and activate or deactivate them together. The buttons in the bar at the
top of the window provide the following options:
Activate selection
Activate all
Deactivate selection
Deactivate all per ring test

quo data GmbH

.

ProLab Plus – Manual Data selection

83

The current assignment table can be viewed as a report by clicking the Report button or
saved as an Excel spreadsheet by simultaneously pressing the “Shift” key .
It is also possible to base your selection on criteria in such a way as to include in your assessment only those results that correspond to specific sample preparations or analytical
methods. Statistical evaluations and the presentation of the evaluation results can then be
generated separately for each sample preparation method and/or each analytical method. To
do this, select the sample preparation method or analytical method and click the Plus button
to include in your selection all the corresponding laboratories, or click the Minus button if the
corresponding laboratories should be excluded. The selection of laboratories according to
categories (as defined in the Basic tables, see 7.2.5) can be carried out as well.
The various ways of activating and deactivating selections which have been described can
be repeated and combined at will.
Please note that selection by categories affects all samples and measurands, irrespective of
which sample or measurand is currently selected for display.
The final selection should be saved in order to be able to make sense of the resulting statistical analysis.

9.4

Save and load selection

If various selections ( e.g. for alternative complementary assessments) have been made, the
information about which selections serve as data basis for which computational steps should
be saved for future reference in order to be able to make sense of the corresponding statistical analysis. That is done in the menu Selection – Save selection.
A maximum of four selections can be saved. In order to save a selection, mark one of the
four storage positions (Selection 1 to 4), enter a title and, if you wish, a description for your
selection. If another selection is already saved in the storage position, it will be overwritten.
Click on Save selection to complete the procedure.

Figure 59: Selection – Load selection

Loading previously saved selections is done under Selection – Load selection. In the window
all available selections with the respective description are shown. Select the desired selection and click on the Apply selection button.

quo data GmbH

.

ProLab Plus – Manual Statistics

84

10 Statistics
10.1 Overview
The statistical analyses are divided into two steps. First, in the analysis (properly speaking),
various ring test parameters such as the mean value and the standard deviation are computed on the basis of various methods. Second, an individual evaluation via Z scores (standardised measurement deviations) based on the previously determined assigned value and target
standard deviation is carried out. The latter two parameters are not necessarily identical to
the (overall) mean value and the (reproducibility) standard deviation, but can be determined
in an intermediate step on the basis of various criteria.
Accordingly, ProLab Plus offers two windows, one for each of the above-mentioned steps.
The following table gives an overview of the procedures pertaining to these windows and of
the correct order for their performance.
Step

No. Function

Analysis of ring
test parameters

1

Determination of estimates for the ring test parameters:
(overall) mean value, (reproducibility) standard deviation, repeatability standard deviation, where applicable: intermediate
standard deviation and/or determination of outliers

2

Determination of assigned value
signed value = overall mean value)

3

Determination of target standard deviation
fault: target s.d. = reproducibility s.d.)

4

Computation of Z scores, Zu scores, Zeta scores, Z’ scores

Computation of Z
scores

(optional; default: as(optional; de-

10.2 Ring test parameters
The aim of the analysis step is to specify the basic ring test parameters such as the overall
mean and the precision measures (repeatability and reproducibility standard deviation). These are then used to determine the assigned value and target standard deviation, the two
quantities on the basis of which the evaluation step is performed. Even if these two quantities
are determined independently of the ring test parameters ( e.g. when reference samples are
used), these ring test parameters must nevertheless be computed so that at least the overall
mean is available for the evaluation step.
Under Statistics – Ring test parameters the following statistical methods are available:
-

ISO 5725-2
DIN 38402 A 42
Swiss Food Manual
Q method and median

-

DIN 38402 A 45 (Q/Hampel)
ISO 13528 / 5725-5 (algorithm A+S)
Nested Design 2x2 (ISO 5725-3/5)
Nested Design (ISO 5725-3)

quo data GmbH

.

ProLab Plus – Manual Statistics

-

85

Q method and Huber estimate

The statistical method according to ISO 5725-3/5 (nested design 2 x 2) is a combination of
the two guidelines. On the one hand, the calculations are performed according to the ISO
5725-3 guideline for nested design, but on the other hand, the rules for robust calculations
according to ISO 5725-5 are implemented. At the moment, this combination is only available
for a 2 x 2 nested design. The statistical method according to ISO 5725-3, however, can be
applied to all nested designs with one factor. The kind of design must have been previously
defined in the Basic tables, and the measurement values must have been entered in a specific order (see section 7.7).

Figure 60: Statistics – Ring test parameter

Having selected a method, click on the Computation button to start the calculation for all selected data.
The evaluation results depend on the calculation method selected. For the statistical methods ISO 5725-2, DIN 38402-A42 and ISO 5725-3 an outlier detection is carried out before
the actual calculation of the ring test parameters. More specifically: the Cochran test is applied for the analysis of variance, and the Grubbs test is applied for the analysis of mean value deviations. For the outlier detection according to ISO 5725-3, a distinction is drawn between the variability of the measurement values within one factor level ( e.g. measurements
performed on one particular day) and between different factor levels ( e.g. measurements
performed on different days). Accordingly, the outlier detection window for the ISO 5725-3
method reflects this distinction by means of the separate displays Cochran test within and
between.
The outlier tests are carried out automatically via the button Detect outlier. A distinction is
drawn between outliers (at the significance level 1 %) and stragglers (at the significance level
5 %). By squares, circles and, additionally, in ISO 5725-3, by diamonds, it is indicated which
of the tests showed a significant result: circle and diamond stand for the Cochran test, while
the square stands for the Grubbs test. If a symbol is completely grey, the corresponding laboratory proved to be an outlier laboratory, if the sign is half white and half grey, the corre-

quo data GmbH

.

ProLab Plus – Manual Statistics

86

sponding laboratory is a straggler.

Figure 61: Legend to ISO 5725-2 (top) and ISO 5725-3 (bottom)

Outlier laboratories are marked by a red cross. The corresponding measurement values will
be excluded from all further calculations. However, it is possible to correct the outlier status
manually with the right mouse button. Furthermore, for the outlier detection according to ISO
5725-2 and ISO 5725-3 the outlier detection tests may be repeated, whereas according to
DIN 38402 A42 this is not allowed and, hence, not possible in ProLab Plus.
The example below shows the results of the outlier detection according to ISO 5725-2 after
activating the button Detect outliers. Laboratory L003 is an outlier according to Cochran, i. e.
quo data GmbH

.

ProLab Plus – Manual Statistics

87

with a significantly differing mean value. It is marked in red in the chart. Laboratory L025 is a
straggler according to Cochran. Only values of laboratory L003 will be excluded from further
calculations.
If you close the window immediately after the calculations have been completed, the empirical values for the mean value and the reproducibility s.d. will be used for the assigned value
and the target standard deviation, respectively.

Figure 62: Ring test parameters – ISO 5725-2

At this point a check for outliers of type E is not carried out. Outliers of type E are defined as
values that lie outside the tolerance limits which are calculated during the Z score computation. Outliers of type E will never automatically be eliminated from the calculation of ring test
parameters by ProLab, as they are not classical outliers.
10.2.1 Option for DIN 38402 A 45: logarithmic calculation
If the box for Logarithmic calculation is checked, ProLab Pluscalculates the mean, standard
deviations and tolerance limits using the logarithmized test results, but always reverts to nonlogarithmic values for the display of results. Moreover, the option is added to the method
name, e.g. in report 34 (Charts and tables – Further reports – Tolerance limits, methods of
analysis and results). Please note that relative standard deviations are meaningless when
computed on the basis of logarithmic data, and that they are accordingly automatically converted using the following formula:
approximate absolute standard deviation of non-logarithmic data
= calculated standard deviation based on logarithmized test results * mean value of
the nonlogarithmic data

quo data GmbH

.

ProLab Plus – Manual Statistics

88

If you prefer a logarithmic scaling of the axes in charts, e.g. in the Summary results, you can
adjust the settings in the chart editor (right click on chart area, menu item Axis).
Please note that the only score-type that can be computed when using logarithmic calculation are Z scores.

An overview of the measurements, laboratory means and standard deviations or of the evaluation results can be obtained by clicking on one of the report buttons for either laboratory
data or results.
After finishing the calculations the assigned value and target standard deviation need to be
defined. If no further data is input, ProLab Plus uses the empirically determined values
(overall mean and reproducibility standard deviation). If you wish to use a different definition
for these quantities, e.g. reference value and the standard deviation according to Horwitz
(Horwitz s.d.), click the buttons for Assigned value and target s.d.
10.2.2 Determine assigned value
Activating the button Assigned value opens a window where the assigned value may be defined for each combination of sample and measurand.
You can choose between the following options for determining the assigned value:
R = reference mean

the reference mean value as specified under Characteristics of
measurands

M = mean

the empirically determined value without considering results below
the detection limit and the limit of determination

lM = lower mean

like M = mean, except that all values below the detection limit and
the limit of determination are set to zero

uM = upper mean

like M = mean, except that all values below the detection limit and
the limit of determination are set to the median of these limits

Mo = mode

main mode of kernel density estimation

Ma = manually

the desired value is entered manually

Please note that by selecting M = mean the assigned valued is defined as the empirical
mean value as computed in accordance with the selected statistical method. If for example
ISO 5725-2 has been selected, the mean value is the arithmetic mean excluding outliers,
while for Q method and median the mean value is defined as the median.

quo data GmbH

.

ProLab Plus – Manual Statistics

89

Figure 63: Statistics – Ring test parameter – Assigned value

In the column Mode the current selection, i. e. one of the possibilities R, M, lM, uM, Mo or Ma
is displayed. The default setting is M. To change the mode, click on the mode field of the relevant row, and select the desired assigned value definition. If you choose the option manually, you can change the value in the Assigned Value column.
To change the mode of all combinations at the same time, open the drop-down menu on the
upper toolbar and select one of the options. The change will become effective upon clicking
on the red check mark.

Figure 64: Determination of assigned value

quo data GmbH

.

ProLab Plus – Manual Statistics

90

10.2.3 Determine target s.d.
Activating the button Target s.d. opens a window where the target standard deviation value
can be defined for each combination of sample and measurand.
You can choose between displaying the relative or absolute standard deviations (as a reminder: the relative s. d. is defined as follows: (s. d. / mean value) x 100). Furthermore, the
following options are available for the definition of the target standard deviation:
R = reference s.d.

the reference standard deviation as specified under Database Characteristics of measurands

L = limited s.d.

if the empirical standard deviation is greater than the upper limit
(assigned value * rel. s.d. max / 100) then the target s.d. is set
to this upper limit; if the empirical standard deviation is less
than the lower limit (assigned value * rel. s.d. min / 100) then
the target s.d. is set to this lower limit; otherwise the target s.d.
is set equal to the reproducibility s.d. Use the option L in order
to prescribe bounds for the range of acceptable empirical
standard deviations, e.g. to prevent the target standard deviation from varying too much from one ring test to the next.. If the
range of tolerance is defined by |Z| < 2 and all results in a range
of +/- 10% of the mean are to be accepted, the value of rel. s.d.
min must be 5. The values of rel. s.d. min and rel. s.d. max are
specified under Database – Characteristics of measurands.

S = reproducibility s.d.

empirical s.d. (the empirically determined reproducibility standard deviation without considering results below the detection
limit and the limit of determination)

eH = empirical Horwitz

empirical Horwitz (standard deviation determined on the basis
of an empirically adapted Horwitz function; the assigned value
established in the previous step is used for the calculation; the
use of an empirically adapted Horwitz function is only recommended if at least three or four different levels have been analysed)

kH = classical Horwitz

classical Horwitz (standard deviation determined on the basis
of the Horwitz function; the assigned value established in the
previous step is used for the calculation)

tH = truncated Horwitz

standard deviation determined on the basis of the Horwitz function; the assigned value established in the previous step is
used for the calculation; if the relative Horwitz-s.d. is greater
than 22 %, the target standard deviation is set to 22 %.

Ma = manually

if this option is selected, any value can be entered in the target
s.d. field. That value may be a relative or absolute figure, depending on the view with which you are working in the table.

quo data GmbH

.

ProLab Plus – Manual Statistics

91

Vf = variance function

the target s.d. is defined as the result of the calculation of the
variance function

SM = s.d. of lab means

standard deviation of the laboratories’ mean values

Figure 65: Statistics – Ring test parameters – Target s.d.

In the column Mode the current selection is shown. The default setting is S, that is the target
s.d. equals the empirical reproducibility s.d.
To select a target s.d. mode for any particular sample-measurand combination click the
Mode field in the relevant row to open a drop-down list box with the options. To change the
mode of all combinations at the same time, open the drop-down list box in the upper part of
the window and select one of the options. The mode change will become effective upon
clicking on the check mark.
10.2.4 Variance function
If you want to use the variance function according to DIN 38402 A 45 as standard deviation,
it has to be calculated beforehand.
By clicking on the Variance function button, the method described below is used to define the
reproducibility standard deviation as a function of the concentration and to check that this
function is sufficiently precise. The method can be used if interlaboratory test results are
available for at least four samples with different concentrations per measurand.
Step 1: Identification of gross outliers – gross outliers can be identified when 4 to 15 samples
are tested
Step 2: Determination of variance function – after the gross outliers have been eliminated,
the final variance function is determined.
quo data GmbH

.

ProLab Plus – Manual Statistics

92

Step 3: Testing the variance function for adequate precision – it may be determined that the
variance function is sufficiently precise. If this is not the case, it is checked whether other factors exert an influence on the variance or whether the functional relationship is more complex.
Step 4: Testing the concentration dependence for significance.

10.3 Computation of Z-scores
This dialogue box allows you to calculate Z, Zu, Zeta and Z’ scores and to assess the laboratories’ performances according to LAWA, BAM or CPCB protocols, or indeed according to
any tailor-made prescription for evaluation. It is found under Statistics – Computation of Z
scores.
The following tabs are available:
Computation
Allows the calculation of Z, Zu, Zeta and Z’ scores as well as the
specification of tolerance limits
Score values

Shows a table with the scores

Results

Shows the evaluation results as well as all Z scores of a selected laboratory

Before the scores can be calculated, the laboratories’ mean values, assigned value and target standard deviation have to be determined first in the Ring test parameter dialogue box.
Additionally it must be ensured that the target standard deviation is greater than zero and for
the calculation of the Zu scores, the assigned value should be greater than zero as well.
The Score values and Results tabs can be opened regardless whether or not a calculation of
the scores has been performed immediately beforehand. The content always refers to
whichever calculation was performed last.
10.3.1 Computation tab
This tab allows you to make your selection for the different pre-settings of the scores and tolerance limits. The table lists the combinations that have been included in the calculations.
The tolerance limits can be set for all combinations at the same time by selecting the desired
option (2, 3 or freely chosen). The tolerance limit for an individual sample-measurand combination can also be entered directly in the column Limit of tolerance in the table. In this case,
the option for each s.-m. comb. individually must be selected.
In accordance with a general trend, ProLab Plus proceeds as follows in the case of borderline scores. If the corresponding rounded scores lie within the range of tolerance, scores lying outside the range of tolerance are subsequently adjusted so as to agree with the rounded
values.
Please note that the number format of measurement values as specified in the Settings or
the Basic tables is applied to the tolerance limits as well.
If for a particular laboratory only values below the limit of determination are available, no laboratory mean value can be determined and thus assessment using Z scores cannot be carried out. If an assessment is nevertheless required, it is possible to automatically assign such

quo data GmbH

.

ProLab Plus – Manual Statistics

93

laboratories a fixed score equal to the lower tolerance limit. In order to do activate this option,
check the box Inclusion of NN and NB results.

Figure 66: Statistics – Computation of scores – Calculation

Z scores:
In order to describe measurement deviations independently of sample and measurand, Z
scores are computed:
Z

test result  assigned value
target s.d.

A disadvantage of Z scores for high relative standard deviations is the fact that laboratories
with systematically lower recovery rates (ratio of the observed to the reference value) are
quo data GmbH

.

ProLab Plus – Manual Statistics

94

privileged. This is clear in the following extreme case: if the relative standard deviation is
greater than 0,5 (s > 0.5m) then even a concentration of zero would be above the lower tolerance limit of -2.
Zu scores:
The tolerance limits for Zu scores must lie between 0.5 and 3.5. In comparison to Z scores
the tolerance limits are raised for the Zu scores. The Zu score has been conceived in such a
way as to ensure that a changing mean recovery rate will not result in any advantages for a
laboratory and that the probability of exceeding the tolerance limits is equal to that for Z
scores.
Zeta scores:
Zeta scores are obtained when the target standard deviation is determined on the basis of
the measurement uncertainty of each individual laboratory.
The measurement uncertainty of each lab has to be specified when entering the measurement values or else no Zeta scores can be determined (see chapter 7.7).
The Zeta score is especially high if the laboratory submitted a small measurement uncertainty and vice versa. If the Zeta score lies in the middle range over an extended period of time,
the specified measurement uncertainty fits the measurement results, i. e. the Zeta score can
be used to check if the individual measurement uncertainty is adequate. It cannot however
be used to evaluate the measurement results themselves.
Z’ scores:
If the uncertainty of the assigned value is not too large, the Z‘ scores will almost completely
agree with the Z scores. In this case, the conclusion can be drawn that the uncertainty of the
assigned value is negligible. However, if the uncertainty of the assigned value is large, Z’
scores may differ significantly from the Z scores. The Z scores will be higher than the critical
values ( e.g. 2.0 or 3.0) and thus provide a warning signal while the corresponding Z’ scores
will not exceed the critical values and thus provide no warning signals.
Finally one has to select the mode of evaluation. This does not affect the Z scores, but the
conclusions drawn from them:
LAWA: A laboratory is considered successful with respect to one particular measurand, if the
percentage of Z scores within the tolerance limits is at least 50 %. It is considered successful
for the whole interlaboratory test if
1. at least 80 % of the Z scores are within the limits
2. and at least 80 % of the measurands have Z scores between the tolerance limits.
Criterion 2 is not effective if fewer than 5 measurands have been analysed. In this case a
corresponding message will be shown.
The percentages are calculated on the basis of all samples and measurands for which analyses had to be carried out. This means that the measurands for which results were expected
will be taken into account even if, for whatever reasons, the results were not reported. Please
note that the phrase ‘samples and measurands for which analyses had to be carried out’ re-

quo data GmbH

.

ProLab Plus – Manual Statistics

95

fers here to the assignment of sample-measurand combinations to laboratories as performed
in the Selection menu, not to the data structures saved in the Database menu.
BAM: A laboratory is considered successful for the whole interlaboratory test if at least 80 %
of all Z scores are within the limits. The percentage is computed based on all samples and
measurands for which the laboratory has delivered results.
CPCB: A laboratory is considered successful with respect to a measurand if all corresponding Z scores are within the limits. It is considered successful for the whole interlaboratory test
if at least 80 % of the Z scores are within the limits.
Individual mode of evaluation: A laboratory is considered successful if at least m % of all
scores are within the tolerance limits (percentage of combinations). It is considered successful for one measurand if more than n % of the scores are within the tolerance limits (percentage per measurand). The percentages are calculated on the basis of the samples and
measurands for which test results have been delivered by the laboratories. m and n must be
assigned separately. In order to select one or both modes for assessment, the boxes must
be marked. If they are not activated, the default value for the assessment according to an indiviual mode is 0%.
Parameter-related evaluation: A laboratory has successfully measured one measurand if
more than 50% of the scores for the corresponding measurand are within the tolerance limits. The percentage is based on all samples which had to be measured by the laboratory. An
overall evaluation of the ring test is not provided.6
10.3.2 Score values tab
This tab provides a table view of all calculated score values.
Click the button Report in order to view a list of the calculated scores and to print them. If you
keep Ctrl pressed while clicking onto the Report button, the same list will be displayed as an
EXCEL document. Clicking the button Lab code allows you to toggle between the display of
the laboratory code or name in the column Lab in the report.
Additionally, you can adjust the table using the options above the table in order to show the Z
scores for all ring tests. As default the sorting order is according to ring tests. If you prefer an
unsorted display of the data, you can right-click the button Description which appears in the
dark-grey area above the table once the checkmark at Only show current ring test has been
removed and select Remove from grouping. If the relevant database is quite large, this new
data display may require a lot of time for implementation. In that case it may be useful to remove the checkmark at Order the results of the ring test in foldable groups in order to speed
up the display of the table. If you have selected the display for all ring tests, it is recommend6

Umweltbundesamt: Empfehlung für die Durchführung von Ringversuchen zur Messung chemischer Parameter
und Indikatorparameter zur externen Qualitätskontrolle von Trinkwasseruntersuchungsstellen, Bundesgesundheitsbl - Gesundheitsforsch - Gesundheitsschutz · 46, 1094–1095 (2003) (German Federal Environment Agency:
Suggestions to conduction of ring tests of chemical parameters and indicator parameters for external quality control of stations that analyze drinking water)

quo data GmbH

.

ProLab Plus – Manual Statistics

96

ed not to show all laboratories, samples and measurands at once but to introduce additional
filters by clicking onto the small rectangles with the down-pointing triangles in the column
headers. Filters can be combined and adjusted with help of a complex editor. The editor can
be opened by clicking the Customize button below the table.
By clicking on a column header, the table is sorted according to that column. Should there be
equal values in that column, you can select another column for further sorting by holding the
Shift key  pressed while clicking into the corresponding column header.
Z scores and means outside the tolerance limits are highlighted with a red background. If laboratories have delivered results from repeated measurements, it is possible that Z score
and mean are within the tolerance limits, while single values are not. Such single outlying
values are highlighted with a yellow background if the corresponding columns are shown.

Figure 67: Statistics – Computation of scores – Score values

10.3.3 Results tab
This tab presents the results of the ring test in two tables. The upper table contains the assessment of the laboratories; the lower one shows the laboratory values in a measurandsample matrix. Display and evaluation may vary according to the chosen modes of evaluation.

quo data GmbH

.

ProLab Plus – Manual Statistics

97

Figure 68: Statistics – Computation of Z scores – Results

You may wish to correct the results manually. In this case you can enter a correction factor in
the upper table. The correction factor will virtually reduce (negative factor) or increase (positive factor) the number of measurements within the tolerance limits for the relevant laboratory
and will thus affect the result.
The results can be viewed and printed in four different report formats. To select a report format, click the corresponding button:
Results:

Overview of the results for all laboratories

Certificate:

A certificate is produced for the selected laboratory containing the laboratory
name, the measurands to be analysed in the ring test, the parameters measured by the laboratory and the assessment successful/unsuccessful.

Annex 1:

Annex to the certificate.
The data are shown samplewise for the selected laboratory and comprise
measurand description, measurement unit, assigned value, laboratory value,
standard deviation and Z-score.

Annex 2:

Annex to the certificate.
The data are shown for the selected laboratory. Laboratory values are presented in a measurand-sample matrix. An overall assessment successful/unsuccessful is provided.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

98

11 Charts and tables
Open the Charts and tables menu to view the results of the evaluation and assessment as a
chart or table. The information contained in the presentation of the results is always based on
the last evaluation. If you have, for instance, first evaluated your data according to ISO 5725,
but have then used the Q method for another evaluation, the presentation of your results will
be based on the results of the robust Q method. Please note that if you change the evaluation method, the laboratory assessment (Score calculation) has to be repeated as well.
The presentation will only include the results of those ring tests, samples, measurands and
laboratories which have previously been selected in the Sample/Measurand Selection and
the Sample/Measurand/Laboratory Selection dialogue boxes. If all results of the evaluation
and the assessment are to be displayed, there is no reason to make a selection. If, however,
the results of certain measurands or laboratories are to be suppressed, you can deselect the
respective data in above-mentioned dialogue boxes.
The menu provides the following options:
-

Summary results
Z score survey
Combination scores
Variances & tolerance limits
Youden plot
Histogram & normal plot
Distribution of Z scores

-

Plot of lab means and repeatability s.d.
Kernel density estimator (KDE)
HORRAT trend
PCA analysis
Mandel’s h & k statistics
Laboratory mean values
Further reports

All graphics and tables can be printed directly or saved in MS-Office compatible formats. For
the tables the export formats EXCEL, RTF (Word), HTML, Text or CSV are available.
Charts may also be exported as WMF or BMP as a separate file or stored on the clipboard.
They can then be inserted into WORD documents or other word processing files. Please
note that WMF files are not interpreted by all programs in the same way, so that in certain
circumstances some objects (titles, scales) may be distorted or displaced. Alternatively you
can save any chart as an SVG-file by right-clicking into the chart area and selecting the option Save as…The Scalable Vector Graphics format allows enlargement without loss in quality and thus results in high-quality pictures in modern programs.
All chart settings can be saved by clicking with the right mouse button on the chart and selecting Save chart settings from the menu. Then enter a name for the setting so that it can be
identified later on. Up to four different settings may be saved for each of the results charts.
Loading these settings can be done via the command Load chart setting. The default setting
may be restored via Reset chart.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

99

11.1 Summary results
The Summary results window allows creating graphics and tables that contain all information
about a single sample-measurand combination. This option may be used for test results, but
also for the laboratory mean values or alternatively the Z scores, the laboratory standard deviations, the reference value, the mean value, the confidence interval for the mean value, the
repeatability and reproducibility standard deviation, the tolerance limits and information on
the chosen evaluation method and the determination of the tolerance limits.

Figure 69: Results – Summary results

First the desired ring test - sample - measurand combination must be selected from the list
boxes in the upper part of the window. Only those combinations can be chosen that have already been selected in the Data Selection dialogue box. The results are displayed as a table
or as a graphic.
If the corresponding option was selected under File – Settings, outliers of type E will be highlighted and displayed in a different colour. If that option was not selected, outliers of type E
will not be identified.
There are six display options available:
Mean + s.d.: laboratory mean values + standard deviations per laboratory,
Mean: laboratory mean values as bars, base line equals overall mean,
Test results: single measurement values,
Z scores: Z scores as bars,
Mean (base line=0): laboratory mean values as bars, base line equals zero,
Mean + mu: laboratory mean values + measurement uncertainties (if available) per laboratory.

quo data GmbH

100

.

ProLab Plus – Manual Charts and Tables

The following Figure 70 shows the six display options for the same data.

Figure 70: Display options: Mean + s.d. (upper left), Mean (upper right), Test results (middle left), Z scores
(middle right), Mean (basis=0) (lower left), Mean + mu (lower right)

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

101

Via the right mouse button, the editor for the chart can be opened in order to carry out individual adjustments. Furthermore, up to four chart settings may be saved for later use.
11.1.1 Right side bar
On the right side of the window there are a few buttons which make it possible to modify the
charts according to individual needs.
Chart

Via the button Chart you can specify which components should be
shown in the chart and define their properties. Especially important is
the submenu Chart lines: once they are marked, the lines ( e.g. mean
value, assigned value) are displayed in the chart. If standard deviation
is selected, six horizontal lines will be drawn. Those lines have a distance of +/-1sR, +/-2 sR and +/-3 sR to the assigned value. If a laboratory evaluation has been carried out, the determined tolerance limits
can be displayed as two horizontal lines. The properties of the lines
(colour, style) can be changed under Properties. To adjust the minimum and maximum values of the Y-axis manually choose Min. Y value or Max. Y value. The Chart Editor provides tools for further customization of the graphic.

Reports

Opens the report preview.

Outlier

In views with laboratory means the displaying of outliers can be
turned on or off. If outliers are to be shown, they will be displayed in a
different colour and marked:
A = laboratories with a single outlier
B = deviant laboratory mean
C = excessive laboratory standard deviation
E = Z scores outside tolerance limits (if selected)

Reproducibility s.d.

Click to have the reproducibility s.d. shown as a bar on the left side of
the chart; if the repeatability s.d. and/or intermediate s.d. were calculated, they will be displayed as well.

Confidence interval
Confidence band

Click to display the confidence interval or confidence band in the
chart. A confidence level of 95 % is used.

Values < QL/DL

Display all values smaller than the limit of determination as a triangle.

Y-Scale

Switches the automatic scaling of the Y-axis on or off. This is useful if
the values only differ slightly and for all graphics (combinations) a
similar scaling is preferred.

Min/Max

Values outside of the chart range will be displayed as arrows at the
top or bottom edge of the chart.

Lab code

Toggles between displaying laboratory name or laboratory code.

Comment

Comments can be added at any place in the chart and will appear in
the report as well. Deleting and adjusting the comment can be performed via Format - Comments.

Info Statistics

The statistical method appears in the upper right corner of the chart.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

102

11.1.2 Menu Format
In the main toolbar, under Format you can find the following submenus:
Default settings

All user-specific changes are cancelled.

Horizontal grid lines Switches horizontal gridlines on/off.
Vertical grid lines

Switches vertical gridlines on/off.

Bar width

Sets bar width automatically. User-specific changes via the Editor will
not be considered.

Automatic axis labels

Implements an automatic labelling of the axes, i. e. manual changes
of the labelling will not be considered.

Lab code

Toggles between displaying laboratory code and laboratory name.

Sort by ...

You can choose between sorting by values, laboratory name and laboratory categories as well as by analytical methods and sample
preparation if available.

Analytical method/ For each laboratory the bar is labelled with the corresponding analytiSample preparation cal method or sample preparation.
Chart editor

Opens chart editor.

Lines

Opens the chart line editor that allows to add, remove or format additional horizontal lines for values, such as mean value, s.d. and tolerance limits.

Header lines

Opens the window for the selection and edition of chart headers. For
more details, see below

Color print

Toggles between monochrome and colour print.

Mail merge

Switches mail merge on and off. Using this option allows all samplelaboratory-measurand combinations to be printed or saved as a PDF
file.

Header lines
It is possible to specify which lines to show on the left or the right side of the chart by clicking
into this area. In the header lines window it is possible to change the font, the line heights
and the horizontal position of the right header lines block.
The wording of several header lines can be changed (e.g. intermediate s.d.) from within the
Settings window.
11.1.3 Menu Output
Reports with tables and/or charts can be opened either via the menu Output in the main
menu bar or the Reports button in the Summary results window. In the Output menu you can
find the following commands:
Save

The graphic will be saved in WMF format.

Copy

The graphic is copied to the clipboard in WMF format for further use.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

103

Report preview

In submenus different options for the report are offered. A preview of
what is to be printed is always displayed.

Print

The last selected type of report is printed directly. If mail merge has
been selected, all available combinations will be shown in the report.
Due to the complexity of the reports (graphics, tables) this operation
may take some time.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

104

11.2 Survey of scores
The Survey of scores dialogue box allows creating charts and tables that contain all information on the scores (Z, ZU, Zeta or Z’ scores).
The scores can be presented by laboratory, sample or measurand. Accordingly, a laboratory,
sample or measurand has to be selected from the upper dialogue boxes. It is also possible to
display all sample/measurand combinations of all laboratories (all combinations).

Figure 71: Charts and Tables – Z score survey (per sample

Only those scores which have previously been calculated under Statistics – Computation of
scores can be displayed.
In the chart, values that lie within the tolerance limits are displayed in blue. Values that lie
outside the tolerance limits are displayed in yellow if they can nevertheless be displayed in
the chart, or in red if they lie outside the range of the chart. In the last case, they are labelled
with their real quantity. The maximum x-value of display for the scores needs to be specified
in the upper part of the window. For example, if tolerance limits of -/+ 2 have been specified
and a maximum x-value of 3, all scores between -2 and +2 are displayed in blue, all scores
between -3 and -2 as well as between +2 and +3 are displayed in yellow and all scores below -3 and above +3 are displayed in red.
In the upper part of the window buttons for a quick modification of the chart settings are
available:
Lab code

Toggles between laboratory name and laboratory code.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

105

Comments

Allows editing and deleting texts that have been included as comments in the chart.

Add. data

States the statistical method applied in the upper right corner of the
chart.

Reports

In submenus, different options for the report are offered. A preview
of what will be printed is always shown.

Additionally in the main menu Format you can find the following submenus:
Default Settings

Cancels all user-specific changes. The Z-score survey window
needs to be closed and opened again after selecting the command.

Horizontal Gridlines

Switches horizontal gridlines on/off.

Bar

Values will be displayed as horizontal bars (as opposed to the default-setting triangles).

Automatic axis labels Implements an automatic labelling of the axes.
Sort by ...

You can choose between sorting by laboratory name, laboratory
code and laboratory categories.

Chart editor

Opens chart editor.

Edit tolerance lines

Change the style of the tolerance lines.

Color print

Toggles between monochrome and colour print.

Description

Toggles between measurand name and code

Comments

Allows editing and deleting texts that have been included as comments in the graphic.

Reports with tables and/or charts can be opened either via the menu Output in the main
menu or the Reports button in the Z-score survey window. In the Output menu you can find
the following commands:
Save

The graphic will be saved in WMF format.

Copy

The graphic is copied to the clipboard in WMF format for further
use.

Report preview

In submenus different options for the report are offered. A preview
of what is to be printed is always displayed.

Print

The last selected type of report is printed directly. If mail merge has
been selected, all available combinations will be shown in the report. Due to the complexity of the reports (graphics, tables) this operation may take some time.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

106

11.3 Combination scores
11.3.1 Chart of RLP and RSZ
The window of the Combination scores allows creating charts of systematic laboratory bias
(RSZ, Rescaled Sum of Z-scores4) and relative laboratory performance (RLP)7. Furthermore,
the coefficient of variation of laboratory (CVL) is shown.
The presentation is either for
one single measurand or all measurands selected and
one single ring test or all selected ring tests.
Furthermore, the time evolution of the scores of one laboratory over several ring tests can be
displayed.
The figures correspond to the last performed calculations under Statistics – Computation of Z
scores and thus to the corresponding selection of score type (Z, Zu, Zeta or Z’ scores).
Interpretation of the combination scores
RSZ is based on a standardized sum of all Z scores (or Zu, Zeta or Z’ scores). This standardization ensures that the RSZ may be interpreted in the same way as a single Z score (or Zu,
Zeta or Z’ score), whereas the difference to the latter is the fact that instead of a single evaluation now an evaluation across samples or measurands or even ring tests is possible. As
long as the RSZ lies within the tolerance limits ( e.g. ± 2), no significant systematic deviations
of the measurement values of the corresponding laboratory can be observed. If the RSZ is
larger than the upper tolerance limit ( e.g. +2), the respective laboratory exhibits values significantly higher than the assigned value and if the RSZ is less than the lower tolerance limit (
e.g. -2), the respective laboratory exhibits values significantly lower than the assigned value.
RLP is the mean length of the Z scores (or Zu, Zeta or Z’ scores) of one laboratory. According to the Harmonized Protocol the RLP is derived from the sum of the squared mean
lengths of all Z scores (or Zu, Zeta or Z’ scores). If a RLP lies close to 1, the respective laboratory exhibits an average performance. A small RLP corresponds to a high laboratory performance, while a high RLP corresponds to a low laboratory performance. If the RLP equals
0.5, the deviations of the laboratory are only about 50% of those of an average laboratory.
CVL is the relative standard deviation of a laboratory. The CVL is determined separately for
each measurand.

7

Uhlig, S., Lischer P.: Statistically-based performance characteristics in laboratory performance studies. The
Analyst 123 (1998) 167-172

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

107

Figure 72: Charts and tables – Combination scores – Table of individual laboratory scores

In the chart, the RSZ values are plotted against the RLP values for the selected ring testmeasurand combination: the bottom axis represents the RSZ values, while the left axis displays RLP values. The rectangle marks the tolerance limits, systematic deviations (RSZ) are
accepted as long as they are not statistically significant, i. e. as long as they are within the
tolerance limits of +/- 2. Deviations in the relative laboratory performance (RLP) are accepted
as long as the mean deviation of a laboratory does not exceed 1.5 times the average deviation of all laboratories. Laboratories with a RLP larger than 1.5 exhibit a variability that is
considerably higher than the reproducibility standard deviation.
Please note that if the scores over several ring tests are to be displayed, each laboratory
needs to be denoted in the same manner in all the relevant ring tests. If laboratory codes are
to be displayed in the chart, the codes of the currently selected ring test (in the main menu)
are used, as laboratory codes can differ from ring test to ring test.
In the example in Figure 73 the RSZ and the RLP values for one measurand (As) in one ring
test (1998_1) are shown. Laboratory 7 exhibits a low RSZ value, whereas the RLP of this laboratory lies within the tolerance limits. Laboratory 137 exhibits a high RLP value and a RSZ
outside the tolerance limits. Laboratory 49 exhibits a RSZ within the tolerance limits but a
RLP value outside the tolerance limits.

Missing laboratory means
In some situations you may want to assign Z scores to non-quantitative test results. For example if the laboratory reported only results below the determination limit for a certain sample-measurand combination, no laboratory mean is calculated. In such a situation, a negative
Z score of -3 may be generated (see section 10.3.1). These scores will not be included in the
RLP and RSZ charts of this window. Instead, an additional column containing the number of
scores included in the calculation is shown in the table on the left. Additionally, there is the
possibility to highlight the ratio of quantitative results by circle diameter or fill color (see the
options in the right-hand sidebar).

quo data GmbH

108

.

ProLab Plus – Manual Charts and Tables

Figure 73: Charts and tables – Combination scores – Chart of RSZ and RLP (measurand As, ring test
1998_1)

In Figure 74, the scores for the same measurand (As) are shown for 4 ring tests (1998_1 to
1998_4). For laboratory 7 now the RLP lies on the upper limit of the RLP, the RSZ is still
within the tolerance limits. Laboratory 49 still differs significantly regarding the systematic deviations as well as regarding the performance.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

109

Figure 74: Charts and tables – Combination scores – Chart of RSZ and RLP (measurand As, all ring tests
of 1998: 1998_1, 1998_2, 1998_3, 1998_4)

Besides the analysis of the scores of all laboratories, the time evolution of the scores of one
laboratory can be also analysed. To do so, mark the checkbox Ring test separately (per lab)
on the right side of the window. The laboratory that should be analysed has to be selected
from the table below the chart by marking the relevant row. Ring test and measurand are selected on the right side.
There are two display options: (1) plotting RSZ against RLP values as in the chart over all laboratories (RSZ vs. RLP) or (2) separate display of RSZ and RLP values (time line).
In the example of Figure 75 it can clearly be seen that in the first ring test (1998_1) laboratory 7 exhibits an RSZ value outside the tolerance limits. In the second and third ring tests
quo data GmbH

110

.

ProLab Plus – Manual Charts and Tables

(1998_2, 1998_3) RLP and RSZ lie within the tolerance limits. In the last ring test of 1998
(1998_4), the laboratory exhibits a RLP value much too high and a RSZ which is close to the
upper tolerance limit.

Figure 75: Charts and tables – Combination scores – Ring tests separately (per lab): RSZ vs. RLP (lab 7,
measurand As, all ring tests of 1998)

Figure 76 shows the time line for the same example as in Figure 75. The RLP value decreases in the first three ring tests, but increases noticeably in the last ring test. The RSZ
value increases steadily for the all four ring tests.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

111

Figure 76: Charts and tables – Combination scores – Ring tests separately (per lab): time line (lab 7,
measurand As, all ring tests of 1998)

In the following example the development of three laboratories over all ring tests between
1997 and 2006 is shown. For the laboratory in the top left corner only the values of two of the
ring tests lie within the tolerance limits, the values of the majority of the ring tests lie above
the RSZ and also above the RLP tolerance limits. Therefore, this laboratory submitted values
significantly larger than the assigned value and also exhibited a high variability of the measurement values in almost all ring tests.
For the laboratory in the top right corner all values over all ring tests lie within the tolerance
quo data GmbH

.

ProLab Plus – Manual Charts and Tables

112

limits. Finally, for the laboratory shown in the bottom figure the majority of the values lie within the tolerance limits. For the values outside the tolerance limits no clear trend can be observed, some of the RSZ values lie above the upper limit, some below the lower limit.

Figure 77: Charts and tables – Combination scores – Ring tests separately (per lab): RSZ vs. RLP

11.3.2 Laboratory assessment based on RLP
Results per measurand & lab history
This tab consists of three parts, which are, from top to bottom:
 a list of selected measurands for the current ring test, grouped by their category 1 as
defined in the Basic tables
 a table showing all labs for which an RLP has been computed for the currently selected measurand. In order for this to be the case, Z scores must have been calculated
beforehand. More information about the table columns is given below
 the performance history plot for the current laboratory
The columns of the table contain the following data:
Name

Laboratory ID (corresponds to the Name entered in Basic tables).

Description

Shows the description entered in Basic tables.

Lab code

Code for the current round as assigned via Database – Encoding.

Rounds

Number of ring tests in which the laboratory analysed the relevant measur-

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

113

and, up to the currently selected round. Only ring tests in Edit mode are considered.
RLP

RLP for the current round. Please note that laboratories for which no measurement data are available will not be shown in the list. Colours are used as
follows to highlight the quality of the result: green if RLP < 1.5, red if RLP > 2,
yellow otherwise.

RRLP

The laboratory’s rolling RLP average, calculated as the mean of the best 4
RLPs taken from the current and its 4 preceding ring tests. Please be aware
that no RRLP can be calculated for the first three active (i. e. Edit-mode) ring
tests. Thus, in order to display an RRLP for a ring test, the four ring tests before the current one need to be made accessible (Edit mode) and the ring
test parameters have to be calculated for those ring tests as well.

P value

A P value below 0.05 is considered an indication for a systematic bias in the
test results.

By clicking on a column header the data is sorted with respect to the corresponding criterion.
A double-click on any laboratory’s row will open the report for that laboratory and the currently selected measurand. Right-clicking on the table allows copying to the clipboard for re-use
in applications such as Microsoft Excel.

quo data GmbH

114

.

ProLab Plus – Manual Charts and Tables

Figure 78: Results per analyte

On the bottom of the window, the performance plot for the current laboratory and measurand
is shown. This chart plots a maximum of three curves:
1. the RLPs for all active (i. e. Edit-mode) rounds
2. the RRLPs for all active ring tests and
3. the median of the RLPs of all laboratories
Again be aware that no RRLPs can be calculated for the first three active ring tests.
The following options are available on the left of the chart:
Hide rounds without RRLP

This option will hide all ring tests for which a RRLP is available for no laboratory.

Show future rounds

This option is only available if the current ring test is not the
last.

Show date below round

Displays the Date of analysis as defined in the Basic tables
below each ring test on the X-axis. Please note that in order for that option to be meaningful, a date needs to be entered for each ring test. As the date text needs more space

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

115

than the ring test number, some ring test labels might disappear but the data points will remain in the chart.
Number format

Change the format of the date used. The default is
yyyy"/"mmm. The following codes can be freely combined:
Code Meaning
yy
yyyy
"/"
m
mm

two-digit year
four-digit year
single slash
month as number
same as m, but with leading
zero for single-digit months
mmm abbreviated month
mmmm full month name
ddddd full date, e.g. 29/05/2010
The last three settings depend on the configuration of Microsoft Windows within the control panel.
The chart options will also be used in the report. With a right-click, more functions for the
chart are offered. Double-clicking the chart will open an extended chart editor.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

116

Results per lab & kernel density estimation

Figure 79: Results per lab

The second tab contains, from top to bottom:
 a selection box for the laboratory of interest. In this box, all laboratories that took part
in the current ring test are shown. They are sorted according to their unique laboratory name (as entered under Database – Basic tables).
 a table of measurands, the last four columns containing the same information as the
corresponding columns of the table in the first tab
 a chart with the kernel density estimation (KDE) of the distribution of the standardised
test results (measured values divided by assigned value) of all laboratories is shown
for the selected measurand. For each sample, separate kernel densities are shown in
different colours. The kernel density over all four test samples is displayed as a thick
black curve. The individual standardized results of all laboratories are shown as circles. The colour of the circle coincides with the colour of the curves. The circles can
be displayed on the overall curve or on the individual curves, depending on which of
the options on the left has been selected. Furthermore, by checking the option Current laboratory only, it is possible to display the standardized results only for the currently selected laboratory rather than for all laboratories.
quo data GmbH

.

ProLab Plus – Manual Charts and Tables

117

These settings are then applied in the report as well. With a right-click in the chart, further options can be selected. Double-clicking the chart will open an advanced chart editor.
Summary of results
On the last tab, the results submitted by the laboratories, the assigned value and the standardised results (ratio of result and assigned value) are shown for each laboratory, each
measurand and each sample of the current ring test (see Figure 80).

Figure 80 – Standardised values tab with popup menu. A filter button is highlighted with a red circle

At the top of the window, limits for the standardized values can be specified. Values below
the lower limit and above the upper limit are highlighted in red . The limits are not used for
any calculation, but only serve as a display option. Check the box on the left of the range to
display only standardised results lying outside the limits (i. e. the ones that are marked in red
already).
The table may be grouped by measurand and laboratory by checking the relevant box above
the table. Once the table has been grouped, a collapsible bar is displayed above each group.
This bar also contains the average standardised value within the group. If the table is not

quo data GmbH

118

.

ProLab Plus – Manual Charts and Tables

grouped, the list can easily be sorted by clicking on the text of any column header. A small
triangle will indicate the sorting direction.
In some cases, the current selection does not contain all laboratories that took part in the
current ring test. If the selection should be taken into account in this table, please check the
relevant box above the table.
It is also possible to filter the data within the table. For example, click on the small arrow button highlighted in Figure 80 to restrict the view to just one laboratory. To remove the filter,
click the [x] button at the bottom left of the table. Filters can also be combined.
A click with the right mouse button on the table opens a menu that allows copying it to the
clipboard or saving it in various file formats. A right-click on any column header opens a different menu to customize the way data is displayed.
Reports
The certificate-like report contains the results of one laboratory for one measurand on one
page. This report can be displayed either
 for the currently selected combination of laboratory and measurand (This report is also generated by double-clicking on a table row within the Results per Measurand and
Results per lab tabs) or
 for all measurands which the currently selected laboratory has analysed (separate
page for each measurand) or
 for all laboratories that analysed the currently selected measurand (separate page for
each laboratory) or
 for all combinations of laboratory and measurand available in the current ring test
11.3.3 Distribution of Z scores
The Distribution of Z scores window displays a histogram and a scatter plot for selected ring
test-sample-measurand combinations over all samples. The difference with the Histogram
and normal plot chart (section 11.7) is that here you can select more than only one ring test,
measurand and/or laboratory.
First you have to select the ring tests, measurands and laboratories you want to include in
the histogram. This can be done on the Selection tab. The selection is carried out by marking
the corresponding ring tests, samples and laboratories. It is also possible to select or deselect all ring tests, measurands and laboratories. If categories for measurands or laboratories
have been defined, all measurands or laboratories within one category can be selected or
deselected with the buttons “+” or “-“.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

119

Figure 81: Charts & tables – Distribution of Z Scores – Selection

On the Histogram/KDE tab you can see the histogram of the Z scores corresponding to your
selection. The Z scores can be displayed by laboratory, by measurand, by laboratory and
measurand or for all data previously selected.
In the upper part of the window you can determine whether the kernel density estimation
(blue) of the Z score distribution and the standard normal distribution (green) should be displayed as additional information. Furthermore, extreme Z scores smaller than -6 and greater
than +6 can be excluded from the chart.
The x-axis represents the Z scores obtained, while, for the laboratory-wise and laboratoryand-measurand-wise chart options the y-axis represents the relative frequency with which
the selected laboratory obtained the respective Z score; for the remaining two chart options
the y-axis represents the relative number of laboratories that obtained the respective Z score.

quo data GmbH

120

.

ProLab Plus – Manual Charts and Tables

The performance of a laboratory can be assessed according to the percentages given in the
top right corner of the chart. These percentages indicate the relative number of scores between -1 and +1, between -2 and +2, between -3 and +3 and finally between -6 and +6. In
brackets the theoretical values according to the normal distribution are shown. In the following chart the distribution of the Z scores of a laboratory with a good performance is shown.
Only about 3 % of the 38 scores lie outside the tolerance limits of -/+ 2. In fact, these 3 % coincide with one score which is above +3. The remaining 97 % of the scores lie between about
-2 and +2.

Figure 82: Charts & tables – Distribution of Z scores – Histogram/KDE

In the window Scatter different scatter plots on the basis of the Z scores selected on the Selection tab are available. It is possible to choose between displaying Z scores, cumulated Z
scores and relative deviations (in %) of the laboratory mean value from the assigned value.
Furthermore, it is necessary to determine whether laboratories, measurands, ring tests or
samples should be shown on the horizontal axis. It is then possible to specify whether the
values are shown per lab, per measurand, per lab and per measurand or for all data previ-

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

121

ously selected. The options per lab and per measurand are only available if either the ring
tests or the samples are shown on the horizontal axis.
In following figure displays the Z scores of the same laboratory as in Figure 82. According to
this chart, copper in sample 51 can be identified as the source of the Z score above +3.

Figure 83: Charts & tables – Distribution of Z scores – Scatter

In the following Figure 84 several display options of the scatter plot are shown. These are
based on the data of two ring tests with 2 samples each, analysed for 4 measurands by 20
laboratories. Except for the values of laboratory 01, all laboratories submitted consistent data. In the chart Z scores for mercury per lab it can clearly be seen that the Z scores of the laboratories of both ring tests for the same samples are very similar: the red and the blue curve
as well as the green and the orange curve almost lie on each other. Only for laboratory 01
the Z scores differ considerably. This is also pointed out in the next chart “Z scores for lab 01
per measurand”, especially for mercury (HG) large discrepancies can be observed: while the
Z score for sample 01 in the first ring test R3 (blue) lies within the tolerance limits a high Z
scores clearly above the upper tolerance limit can be seen for the same sample in the second ring test R4 (red). Similar trends can be observed for measurands nickel (Ni) and zinc
(Zn). The deterioration of the results of this laboratory from one ring test to the next can also
be seen quite well in the next chart Z scores for lab 01 per ring test:, the Z score for mercury
(green) clearly increases from -1.5 in ring test R3 to +3 in ring test R4. Whereas the chart Z
scores for mercury per ring test shows that only laboratory 01 exhibits such remarkable deviations for the measurand mercury, while the curves for the remaining laboratories are rather
horizontal, i. e. for these laboratories hardly any differences between the Z scores for mercury of both ring tests can be detected. By displaying the Z scores per sample once for labora-

quo data GmbH

122

.

ProLab Plus – Manual Charts and Tables

tory 01 over all measurands and once for mercury over all laboratories, the strong deviations
of the Z scores of laboratory 01 become clearly visible once again.

Figure 84: Charts and tables – Distribution of Z scores – Scatter

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

123

11.3.4 Control chart and extended evaluation
In this window, the control chart is generated and the combination scores per measurand
and for all measurands are calculated. The calculation is based on the parameters assigned
value and target s.d. as defined under Statistics – Ring test parameters.
There is the possibility to cap Z scores, i. e. a capping limit is defined. According to the protocols of AOAC, IUPAC and ISO, a capping limit of 3 is recommended. This means that instead of scores such as -3.3 or -8.9, the score -3 is assigned. Capping is only useful if combination scores are about to be calculated. If there were no capping, a single gross outlier
could distort all combination scores.
After specification of the single scores, the parameters for confidence intervals as well as the
assigned variability limits have to be defined. The laboratory, environmental or consumer risk
( e.g. 0.1) describes the acceptable statistical error probability of a laboratory being wrongly
evaluated, whereas the variability limit ( e.g. 0.2) describes a hypothetical maximum relative
error.
Please note that the calculation of Lischer scores takes a certain amount of time. Accordingly, the calculation of these scores can be activated separately.
On the tab Control chart you can find a graphic display of combination scores for all measurands. These can be shown either lab-wise or for all laboratories.

11.4 Test on equivalence
11.4.1 Introduction
The principle of equivalence states that two analytical methods have to be considered as
equivalent if their deviation falls below a certain value to a certain level of statistical significance. The proof of equivalence always comprises several aspects. Thus in the context of a
ring test it has to be ensured that besides the deviation of the general mean (recovery equivalence), the standard deviation under repeatability (repeatability equivalence) and reproducibility (reproducibility equivalence) conditions do not exceed a given value. The greater the
number of replicates of measurements performed, the sharper the conclusions on equivalence can be formulated, from a statistical point of view. If there are only a few replicates,
equivalence can only be proved if relatively large deviations are accepted.
Equivalence can a priori only be examined for one single sample. However, measurements
of different samples can be considered together, as long as it is ascertained that samplespecific method effects are negligible. Otherwise it cannot be excluded that the result does
not reflect the efficiency of the analytical method in general, but may be influenced by the
particular sample selection. Regarding recovery equivalence, this can be determined by calculating the difference between the sample specific and the average relative deviation. If the
confidence interval calculated for the variances of this difference covers the zero point for all
samples, it can be concluded that sample-specific method effects are not provable. You can

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

124

find a corresponding chart in the window Charts and tables – Test on equivalence – Sample
specific method effects. (see also section 11.4.6)
11.4.2 Define classes of methods
Basically every combination of measurand and method should undergo an independent test
on equivalence. Here, method denotes the combination of sample preparation and analytical
methods. For practical reasons ( e.g. insufficient data) it may be reasonable to combine several methods. Of course this is only permitted if it can be assumed that the methods are
equivalent. The calculation of equivalence is then carried out for every combination index, no
matter whether it contains only one method or a combination of equivalent methods.
At first the classes of methods have to be defined. This can be done in the window Charts
and tables – Test on equivalence – Classes of methods. The currently defined classes and
the respective colour codes are displayed on the bottom side of the window. To add, edit or
delete classes of methods, click on the Edit class of method button.

Figure 85: Test on equivalence – Edit classes of methods

The assignments of the classes to the single methods must then be carried out. To do this,
select the relevant class on the bottom. Double clicking on a cell of the method/measurand
table allocates the chosen class and colours the cell respectively. A repeated double click
deletes the allocation. For these operations you can also use the buttons in the above menu:
Add class reference,
Delete class reference. You can also select several cells and
assign them to one method class together..

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

125

Figure 86: Statistics – Test on equivalence – Classes of methods

You can choose between displaying all measurands available in the database or only the
measurands belonging to Category 1 or Category 2.
A summary of the available data of the current allocation is shown at the bottom of the Classes of methods table.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

126

11.4.3 Recovery equivalence
Equivalence with respect to recovery means that the relative difference of the overall mean
values of the reference method and the method being evaluated falls below a particular tolerance value  Recovery . Typical values for  Recovery are in the range of 10-20%. For the level of
significance  usually a value of 5%, 10% or 15% is defined. Click on the button Start to carry out the calculations using the current settings.

Figure 87: Statistics – Test on equivalence – Recovery equivalence

Besides the general statistical parameters for the analyses of the samples by different methods, the parameters of the determination of equivalence are also displayed:
Non-centrality parameter of the assumed non-central t-distribution.
Critical value that is derived from the non-central t-distribution for the specified parameters and the non-centrality parameter.
On the basis of the preceding, the Maximal tolerated empirical deviation can now be
computed.

If the actual Empirical deviation is smaller than the maximal tolerated empirical deviation, the equivalence for the specified tolerance  Recovery is proved at the level of significance 

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

127

11.4.4 Reproducibility equivalence
Two methods are equivalent with respect to their reproducibility standard deviations, if the
latter differ by less than a tolerated factor  R . Appropriate values for the maximal tolerated
ratio  R are around 1.6.

Figure 88: Statistics – Test on equivalence – Reproducibility equivalence

The statistical parameters for the reproducibility standard deviation of both methods are displayed as well as the
Standard deviation of the empirical difference,
the (1-)-quantile of the standard normal distribution,
-

from which the Maximal tolerated empirical difference for the specified  R and  is



computed.
If the actual Empirical difference is smaller than the maximal tolerated empirical difference, the equivalence for the specified ratio  R is proved at the level of significance 

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

128

11.4.5 Repeatability equivalence
The equivalence of two methods concerning the repeatability standard deviation is computed
analogously to the reproducibility standard deviation.
Accordingly, the values of  r should also lie around 1.6. The output is analogous to that of
the reproducibility standard deviation.

Figure 89: Statistics – Test on equivalence – Repeatability equivalence

11.4.6 Sample specific method effects
As explained in the introduction (see section11.4.1), an evaluation over all samples is only
permitted when the methods show no sample-specific effects. In the chart Sample specific
method effects, the confidence intervals for the differences of the sample specific to the average relative deviation are shown. If all of them cross the zero line, no significant samplespecific method effects can be detected.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

129

Figure 90: Statistics – Test on equivalence – Sample specific method effects

11.4.7 Survey of methods
A survey of all methods for every measurand is shown on the tab Survey of methods.
11.4.8 Export of results
The results shown in the various tables can be exported in several formats. By clicking with
the right mouse button in a table, the following context menu can be opened:
Copy
Copy selection

Copies the whole table to the clipboard.
Copies the marked selection of the table to the clipboard.
Please note that this can be pasted into an Excel table,
but not into a Word document.
Save in file
Saves the table as a text file.
Copy to Excel
Saves the whole table as an Excel table. Excel must be
available on the PC
Set column width
Enter a value to set the column width.
Adapt column width automatically Adapts the column width automatically.
Line break
Line break in column header if necessary.
Select column
Selection of columns to be shown in the table

11.5 Tolerance limits and rel. standard deviations
In this window you see two charts that represent the tolerance limits, variances and standard
deviations on the basis of the assigned value.
Buttons for a quick adjustment of the chart settings are available in the upper part of the window:

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

130

Comment

Allows editing and deleting texts that have been included as comments in the
chart.

Report

In submenus different options for the report are offered. A preview of what is
to be printed is always shown.

Figure 91: Charts and Tables – Tolerance limits and rel. standard deviations

Further information on the values displayed in the
chart can be obtained by moving the mouse pointer
close to the value. The pointer takes on the shape of
a hand. If you now press the left mouse button, a
window with the corresponding laboratory name,
assigned value and relative standard deviation
opens up.

11.6 Youden plot
The Youden plot window displays a combined graphic of the results of two samplemeasurand combinations. Such a presentation allows identifying systematic effects in the laboratory-specific deviations.
The following combinations are possible for the two axes:

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

-

-

131

per measurand: the two axes represent two different samples for the selected measurand
across measurands: all measurands of the two samples are combined
per sample: the two axes represent two different measurands for the selected sample
across samples: all samples with the two measurands are combined

Depending on the combination you have to choose the common measurand or sample and
the desired measurands and samples for the axes via the scroll-down menus on the right.
For the combinations over all measurands and over all samples the scroll-down menu for
common measurands and common samples are not displayed.

Figure 92: Charts and tables – Youden plot – Zu scores (per measurand)

quo data GmbH

132

.

ProLab Plus – Manual Charts and Tables

Figure 93: Charts and tables – Youden plot – Laboratory mean values (per sample)

The graphic can be printed or edited by using the following buttons:
Code

Toggles between laboratory name and code

Report

Opens a window with a report preview in the currently selected report format

Frame

Toggles between no frame or a frame as square, ellipse or circle. The ellipse is
drawn for a significance level of 5 %.

Limits

Change of the value of the borderline. The value equals the radius of the circle
or half the edge length of the square.

A graphical combination per sample or per measurand is available on the Scatter tab.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

133

Figure 94: Charts and tables – Youden plot – Scatter

11.7 Histogram and normal plot
The Histogram and normal plot window displays the graphics of the histogram and the normal plot for one selected ring test-sample-measurand combination.
The charts are displayed on separate tabs. Click on the corresponding tab to select a chart.
The toolbar contains buttons for editing or printing the charts as well as drop-down menus for
selecting the ring test, sample and measurand.
With the histogram the underlying statistical distribution of the measurement errors can be
identified visually, and additionally a statistical test (Chi-square-test) for normal distributions
is provided. If the p value presented in the figure is smaller than 0.05 it can be concluded that
the assumption of normal distribution can be denied at the significance level of 5 %.
It has to be noted that the test result is strongly dependent on the possible existence of outliers. Hence it is possible to exclude all measurement values with an absolute value of the Z
score higher than 3 (|Z score|<=3) or – if ISO 5725-2 or DIN 38402 A 42 are used – those
measurement that have been previously excluded as outliers (without outliers). This can be
done via the scroll-down menu in the center of the top of the window.
In the following figure the histogram for cadmium in sample 04 is shown, the upper histogram
including all available values and the bottom one including only absolute Z scores less than
quo data GmbH

134

.

ProLab Plus – Manual Charts and Tables

3. In the first case the assumption of normal distribution must be rejected (p value < 0.05),
but not in the second case (p value = 0.951).

Figure 95: Charts & tables – Histogram & normal plot – Histogram: all values (left), only Z scores between
-3 and 3 (right)

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

135

With the normal plot the distribution of the measurement errors can also be surveyed visually. If these are normally distributed, the observed and the expected Z scores must be almost
congruent, i. e. all points have to be approximately on the same line.

quo data GmbH

136

.

ProLab Plus – Manual Charts and Tables

Figure 96: Charts & tables – Histogram & normal plot – Histogram: all values (top), only Z scores between
-3 and 3 (bottom)

11.8 Chart of lab means and repeatability standard deviation
In the window Chart of lab means and repeatability standard deviation, for selected sample-

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

137

measurand-combinations the mean values and standard deviations of each laboratory are
shown. The sample and measurand can be selected in the upper part of the window.
On the right side of the window it is possible to specify whether the tolerance range should
be displayed as rectangle or, as suggested in ISO 13528, as circular areas for the levels of
0.1 %, 1 % and 5 %.

Figure 97: Charts and tables – Chart of repeatability standard deviations

The values of each laboratory with the respective laboratory name or alternatively the laboratory code are displayed as blue circles. The standard deviation pertaining to a particular laboratory can be read off the y-axis, the corresponding mean value off the x-axis.
Laboratories lying outside the tolerance limits exhibit values outside of the confidence interval. Either they have significantly smaller or larger mean values than the average mean value
over all laboratories or a significantly larger standard deviation.
For every selected sample-measurand-combination the y-axis scaling is adjusted. If it is desirable for the scaling to remain constant, e.g. for a better comparison of different combinations, the automatic scaling can be turned off by clicking on Y scale off.
The name of a laboratory can be moved by clicking on the name with the left mouse button
and dragging it to the desired place while holding down the left mouse button. This can be
quo data GmbH

.

ProLab Plus – Manual Charts and Tables

138

useful when many laboratory values lie close to each other or when one value lies close to
the edge of the chart. Such a displacement can be cancelled by refreshing the chart, e.g. by
selecting the sample-measurand-combination anew.
The buttons in the upper part offer functions such as report preview, adding comments and
saving the chart in different styles. To edit the layout of the chart the button Chart settings is
available. Minimum or maximum values for the y-axis can be defined, horizontal and vertical
gridlines can be shown or hidden and an Editor can be opened.

11.9 Kernel density estimator (KDE)
The Kernel density estimator (KDE) window shows the kernel density estimation of the distribution of the laboratory mean values for one selected measurand/sample combination over
all laboratories or – if desired – for one selected laboratory.

Figure 98: Charts & tables – Laboratory performance summary (over all labs)

The dark blue curve is the data distribution calculated employing kernel density estimates. If
instead of the reproducibility standard deviation the Horwitz standard deviation (if computable) should be used as the parameter for the bandwidth, click the corresponding button.
If an outlier detection has been carried out, the dark blue curve presents the distribution of
the measurement values after the outlier elimination, i. e. of the measurement values which
serve as the basis for all statistical analyses. Via the button Outlier, the distribution of the outlier values can be shown as a red curve (if an outlier detection has been carried out). In that
quo data GmbH

.

ProLab Plus – Manual Charts and Tables

139

case the sum of the blue and the red curves (i. e. the sum of the measurement value distribution without outliers and the outlier value distribution) is shown as a black curve. It might be
covered in parts by the red or the blue curve.
Via the button Single values, the single measurement values can be displayed as little circles
on the kernel density curve; red circles indicate outlier values which have been eliminated
from the dataset.
Additionally, the overall mean and – if available – the assigned value and the reference value
are displayed. The overall mean with its corresponding 95 % confidence interval is shown as
a green horizontal bar alongside which the exact values are given. Values outside this interval might be caused by a systematic laboratory error ( e.g. change in test conditions).
The assigned value is displayed as a green triangle, the reference value is displayed as a
blue triangle or if a reference standard deviation is available it is displayed as a blue horizontal bar. Again the exact values are shown next to the symbols. If the assigned value agrees
with the overall mean or with the reference value, it will not be displayed separately.
The tolerance limits, insofar as Z scores have been computed, are shown as two red arrows.
Via the button Number of labs, the empirical distribution of the laboratory mean values is
shown in the chart as a light blue step function with the number of laboratories displayed on
the right axis. Via the button Cumulative Distribution additionally the cumulative distribution
function of the laboratory measurement values can be displayed in the chart.
On the horizontal axis the following ranges around the overall mean are marked in different
colours:
Values that differ less than one time the reproducibility standard deviation from
Green
the assigned value
If no Z scores have been calculated so far:
Values that differ more than one time but less than two times the reproducibility
standard deviation from the assigned value
If Z scores have been already calculated:

Blue

Values that differ more than one time the reproducibility standard deviation from
the assigned value, but lie within the tolerance limits
If no Z scores have been calculated so far:
Values that differ more than two times but less than three times the reproducibility
standard deviation from the assigned value
If Z scores have been already calculated:

Yellow

Values that lie outside of the tolerance limits, but differ less than three times the
reproducibility standard deviation from the assigned value
Values that differ more than three times the reproducibility standard deviation
Red
from the assigned value
Furthermore, the modes (local maxima of the curve) are marked on the curve by arrows and
their exact values are displayed. If there is only one mode, the distribution is called unimodal,
for two modes it is called bimodal. A bimodal distribution indicates that there may be two
populations of laboratories, i. e. that there might be two groups whose results differ noticea-

quo data GmbH

140

.

ProLab Plus – Manual Charts and Tables

bly.
Additionally the parameters of one selected laboratory may be displayed in the chart. To enable this option, mark the checkbox in the upper right part of the window and select the desired laboratory from the scroll down menu. The name of the laboratory is displayed on the
top right of the chart.

Figure 99: Charts & tables – Laboratory performance summary (for one lab)

The black triangle represents the mean value of the selected laboratory, with the corresponding value highlighted in grey above the triangle. Two vertical black bars mark the range laboratory mean value +/- 2 times standard deviation within laboratory. If the assigned value
lies within this range, it can be assumed that the laboratory submitted unbiased data.
For a nested design ( e.g. laboratory reports 4 measurement values, two for the first day and
two for the second day), both mean values are displayed as black crosses on the kernel
density curve, the range mean value +/- standard deviation is shown as a horizontal bar. If
both mean values lie between the two vertical lines, it can be concluded that there is no significant difference between the two days.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

141

11.10 HORRAT trend
The HORRAT trend window displays the HORRAT values (Horwitz Ratio), e.g. the ratio between the empirical and classical Horwitz standard deviations. You can choose between displaying the HORRAT values for one selected measurand or for all measurands.
In order to display the HORRAT values for one single measurand, check per measurand on
the right and select the desired measurand. In the chart the HORRAT trend is plotted as a
function of the concentration. Different ring tests are shown in different colours.

Figure 100: Charts & tables – HORRAT trend (per measurand)

In the across-measurands figure, the HORRAT values over all available measurands and
samples in different ring tests are shown. Different colours indicate different measurands,
while the x-axis shows the sample and the ring test.
If the HORRAT values are displayed for all measurands, it is possible to connect the dots by
clicking on the Line button.

quo data GmbH

142

.

ProLab Plus – Manual Charts and Tables

Figure 101: Charts & tables – HORRAT trend (all measurands)

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

143

11.11 Principal Component Analysis (PCA)
With the help of a Principal Component Analysis it is possible to generate a graphic display
of correlations between measurands. It is furthermore possible to detect specific features that
are distinctive of individual samples or laboratories.
Measurement result deviations from the target value, such as Z scores, are quite often highly
correlated between different measurands, at least when the measurements are performed in
the same step. These correlations are especially interesting when looking for potential
sources of errors. The Principal component analysis provides statistical tools for the systematic analysis of these correlations.
On the tab Measurands you can see a two-dimensional presentation of the similarities of the
measurands. If two measurands are close together, the measurements are highly correlated.
On the other hand, if two measurands are far away from each other, the correlation is very
weak. High correlation between measurands may be used as an indication that the measurements deviations for these measurands are caused by the same source of error.
On the tab Samples/Laboratories you can see a presentation of all samples measured by the
laboratories. If a few of a laboratory’s samples lie close together, this is an indication that the
same kind of error affects this laboratory’s measurements for these samples for all measurands.
The above statements only hold if the principal components are significantly non-zero.

Capping
The PCA of scores will yield less readable results for frequently encounters scores, if there
are extreme scores present. That is why, by default, scores are capped before the principal
components are calculated. This behaviour can be changed from within the Options button.

quo data GmbH

144

.

ProLab Plus – Manual Charts and Tables

Figure 102: Charts & tables – PCA (top: Measurands, bottom: Samples/laboratories)

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

145

11.12 Mandel’s h & k statistics
Mandel's h statistics and Mandel's k statistics present measures for graphically surveying the
consistency of the data. They are helpful for laboratory assessment as well as for describing
the variability of a (harmonized) measurement method.
Mandel's h statistics is used to determine if there are differences between the mean values of
the laboratories, while Mandel’s k statistics is used to evaluate the variance of each laboratory compared to the variances of the other laboratories.
The examination of the plots of Mandel's h and k statistics may indicate that specific laboratories exhibit patterns of results that are markedly different from the others. This is indicated
by consistently high or low variation (compared to the other laboratories) and/or extreme
(whether high or low) mean values.
In the following figures Mandel's h and Mandel's k statistics are shown for each laboratorysample-combination. The legend next to the figure explains the sequence of the bars for
each laboratory: the first entry in the legend coincides with the bar furthest to the left (for one
laboratory), while the last legend entry coincides with the bar furthest to the right (for one laboratory).
Values differing to a high level of statistical significance from values of the other laboratories
are marked in a different colour: a red bar indicates a value significant to the level of 1%
while a yellow bar indicates a value significant to the level of 5%.
Various patterns can appear in the plot of Mandel's h statistics. Each laboratory can have
both positive and negative values. Individual laboratories may tend to have either all positive
or all negative values. This is no unusual pattern, but it may suggest that a common source
of laboratory bias exists.
If you are interested in the absolute values of the h statistics only, mark the checkbox in the
right part of the window. Then significant mean values can still be identified but it is no longer
possible to tell if they lie too high or too low.
In Figure 103, laboratory L048 stands out. This laboratory exhibits significantly higher mean
values for three samples. All other laboratory-sample combinations show no significant deviations.

quo data GmbH

146

.

ProLab Plus – Manual Charts and Tables

Figure 103: Charts & tables: Mandel’s h & k statistics – Mandel’s h statistic

If one laboratory stands out on the k statistics plot as having many large values, it exhibits a
poorer repeatability precision than the other laboratories. A laboratory could have consistently low k values because of such factors as excessive rounding of its data or an insensitive
measurement scale.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

147

Figure 104: Charts & tables: Mandel’s h & k statistics – Mandel’s k statistic

For a nested design, Mandel’s k- statistics is displayed separately for the variability within
one factor level (within) and the variability between the factor levels (between).
In Figure 105, Mandel’s k- statistics for the variability of the measurement values within a factor level is shown. Three outliers can be observed.

quo data GmbH

148

.

ProLab Plus – Manual Charts and Tables

Figure 105: Charts & tables - Mandel’s h & k statistics – Mandel’s k statistic for within standard deviation
for nested design

11.13 Laboratory mean values
Via the menu Charts and Tables – Laboratory mean values a comprehensive summary of all
statistical results either for one sample or for one measurand is offered.
It is possible to choose between the display of laboratory mean values or Z scores. Furthermore several ring test parameters such as overall mean value and standard deviations are
shown. Click on the button Add data to open the dialogue box for the selection of the desired
parameters.
The Limit multiplicator is necessary in order to determine reproducibility limits as well as repeatability and intermediate limits if the respective standard deviations are available.

quo data GmbH

.

ProLab Plus – Manual Charts and Tables

149

Figure 106: Charts & tables – Lab mean values – per measurand

11.14 Further reports
The Charts and Tables – Further reports menu item allows creating more tables with test and
evaluation results.
The following tables are available:
Number of test results per lab: contains the number of test results and participating laboratories for all ring tests (sorted by ring test) and all sample-measurands combinations.
Survey of test results: opens a flexible tabular data window. As is true for most tables within
PROLab, export options are also available in this window. Data fields can be selected by
clicking with the right mouse button into the table. It is possible to choose between a display
of data from all ring tests, from the current ring test or from a selection of ring tests.
To generate a report of the table use the button Report.
Assigned value, target s.d. and tolerances: shows a table with the assigned value and target
standard deviation and their modes as well as the tolerance limits and the statistical method
applied for each sample-measurand combination.
Mean value and standard deviation: shows a table with the reference value, the overall mean
value, the reproducibility s.d., the empirical and classical Horwitz s.d., the repeatability s.d.,

quo data GmbH

150

.

ProLab Plus – Manual Charts and Tables

the tolerance limits, the unit, the number of laboratories and of test results as well as the statistical method applied for each sample-measurand combination.
Tolerance Limits: shows a table with the assigned value and the tolerance for each samplemeasurand combination.
Method parameters: contains a table with the overall mean value, the reproducibility and repeatability standard deviations (both absolute and relative) as well as the reproducibility and
repeatability limits (both absolute and relative) for each sample-measurand combination.
Summary of outliers (current ring test): offers an overview of all eliminated outliers for the
current ring test.
Summary of outliers (labwise): offers an overview of all eliminated outliers for each laboratory.
Own reports: opens a dialogue box where you can open and view a report file saved earlier.

quo data GmbH

.

ProLab Plus – Manual Quick Start: The steps of statistical analysis

151

12 Quick start: The steps of statistical analysis
In order to get a first impression of the statistical core of ProLab Plus, one can perform an
analysis according to the following procedure. This procedure is a simple way to evaluate a
ring test. However, there are many other possibilities, and for the definition of a general procedure many aspects have to be taken into account, such as:
-

the purpose of the ring test
the type of the measurands
the performance level of the laboratories and
the level of heterogeneity of the laboratories

Step 1: Selection of data to be evaluated in the Selection menu
(a) Select sample-measurand combinations to be analysed

Figure 107: Selection – Sample-measurand

quo data GmbH

152

.

ProLab Plus – Manual Quick Start: The steps of statistical analysis

(b) Select/Deselect sample-measurand-laboratory combinations to be analysed

Figure 108: Selection – Sample-measurand-laboratory

Step 2: Statistical Analysis / Assigned value and target s.d.
This step consists of the determination of the basic parameters for the evaluation, the assigned value and the target standard deviation. In subsequent steps these parameters are
used to calculate Z-scores.
(a) Select statistical method to be applied (ISO 5725, Q-method etc.) and start calculation
by clicking the calculator button labelled Computation.

Figure 109: Statistics – Ring test parameter

(b)
In case the assigned value should not be the mean value, click on the button Assigned value and select the appropriate mode.

quo data GmbH

.

ProLab Plus – Manual Quick Start: The steps of statistical analysis

153

(c) In case the target standard deviation should not be the reproducibility standard deviation, click on the button Target s.d. and select the appropriate mode. You should also activate this window if Horwitz s.d. and empirical Horwitz s.d. should be calculated (and
visualized in order to compare with the empirical result)
(d) Click on the report buttons to obtain tables of lab results and ring test results.

Step 3: Statistical Analysis / Calculation of scores
This step consists of the calculation of scores, the evaluation of these scores according to
specific schemes, and the generation of tables of scores and certifications.
(a) Determinate type of score, limit of tolerance (=2) and type of assessment (LAWA, BAM,
CPCB or individual evaluation mode).
Start computation by clicking on the Computation button.

quo data GmbH

154

.

ProLab Plus – Manual Quick Start: The steps of statistical analysis

Figure 110: Statistics – Calculation of scores

(b) Click on Score values to obtain a table of the scores.
(c) Click on Results in order to obtain the evaluation results according to specific evaluation
concepts. This menu presents the results of the ring test in two tables. The upper table
contains the assessment of the laboratories; the lower one shows the laboratory values
in a measurand-sample matrix.

quo data GmbH

.

ProLab Plus – Manual Quick Start: The steps of statistical analysis

155

Figure 111: Statistics – Calculation of scores (Results)

Step 4: Generate reports and charts in the Charts and Tables menu as required.
You may print out the results using the report forms, and in many cases you can also obtain
EXCEL-like presentations. Charts can be copied into the clipboard and then pasted e.g. into
WORD.

quo data GmbH

156

.

ProLab Plus – Manual Some aspects of the statistical methods in ProLab Plus

13 Some aspects of the statistical methods in ProLab Plus
13.1 When can the method of the Swiss food manual be applied?
This method was designed primarily for method validation ring tests, but due to its statistical
robustness it can also be used for the analysis of proficiency testing schemes. However, its
robustness is limited: not more than 20% of the labs may be underperformers in order to
guarantee that the method yields reasonable results. Note that replicates are required for this
method.

13.2 When can Q/median, Q/Huber or DIN 38402 A45 be applied?
These methods are designed primarily for the analysis of proficiency testing schemes, but
they can also be used for method validation ring tests. If the conditions assumed normally for
method validation ring tests hold, Q/median, Q/Huber and DIN 38402 A45 give results that
are very similar to ISO 5725 and DIN 38402 A45. The methods remain viable with up to more
than 40% outlier labs. In that case the robust s.d. will increase by up to 100% (only under
very extreme circumstances); however the results are still reasonable. Note that replicates
are not required for these methods.
Q/median, Q/Huber and DIN 38402 A45 differ with respect to the calculation of the mean
value. Q/median is simply based on the median estimator, whereas Q/Huber applies the statistically more effective Huber estimator (as described in Huber 1981). DIN 38402 A45 uses
the Hampel estimator instead of the Huber estimator.

13.3 When can ISO 5725 and DIN 38402 A42 be applied?
ISO 5725 and DIN 38402 A42 are methods for the statistical analysis of method validation
ring tests, i. e. they should not be used for the analysis of proficiency testing schemes. For
the calculations according to these standards the following specific assumptions are made:
 All labs (apart from only a very few outlier labs) must have equal analytical performance (*clearly this assumption cannot be made for proficiency tests)
 All labs must use the same analytical method (in order to guarantee that distribution
of test results is close to the normal distribution)
Both ISO 5725 and DIN 38402 apply the Grubbs test for the outlier identification of individual
test results and laboratory mean values.
Additionally tests for the identification of excessive intra-laboratory standard deviations are
applied (Cochran test and F-test, respectively).
Note that both methods require replicates.

13.4 When can the Horwitz function be applied?
The Horwitz function does not take different measurands, different samples and matrices,
and different analytical methods into account. It gives an average s.d. for the results of a
large number of ring tests performed over the last 40 years, based only on measurand concentration (according to the statistical analysis of W. Horwitz).
The Horwitz function is only recommended for an assessment of the general performance of

quo data GmbH

.

ProLab Plus – Manual Some aspects of the statistical methods in ProLab Plus

157

the labs and the general outcome of the ring test, but not for the calculation of individual Zscores. Otherwise it may happen that more than 80 % of the Z-scores exceed the tolerance
limits (in case the general performance is far below the average performance in the ring tests
considered by Horwitz).
Horwitz observed that the reproducibility c.v. exceeded twice the Horwitz c.v. only in very few
ring tests, suggesting that these ring tests may be flagged as critical (characterized by problems with the methods or with the ring test samples). This is only a very crude rule of thumb.

13.5 When can the empirical Horwitz function be applied?
The empirical Horwitz function assumes that for each measurand there is a specific Horwitz
type relation between concentration and s.d., but this Horwitz type function is flexible in its
two parameters. Its calculation requires at least two samples and cannot be applied in case
the samples are different with regard to preparation and matrix.
Principally, it can be calculated on the basis of the data from several ring tests, but it is not
recommended to use data spanning more than 3-8 years.

13.6 When can the nested design analysis according to ISO 5725-3 and ISO
5725-5 be applied?
The nested design analysis is required to calculate not only reproducibility and repeatability
standard deviation, but also the intermediate precision standard deviation as described in
ISO 5725-3. Apart from these standard deviations, robust measures for the mean and the
standard deviation of the laboratory averages according to algorithm A of ISO 5725-5 (which
is equivalent to the robust method in ISO 13528) are calculated.

quo data GmbH

158

.

ProLab Plus – Manual Some aspects of the statistical methods in ProLab Plus

14 References
ProLab Plus contains JCL and JVCL source code that can be obtained from:
http://sourceforge.net/projects/jcl/
http:// sourceforge.net/projects/jvcl/

quo data GmbH

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close