Social Inequalities in Education

Published on June 2021 | Categories: Documents | Downloads: 8 | Comments: 0 | Views: 31
of x
Download PDF   Embed   Report

Comments

Content

 

 European Education, 47: 295–310, 2015 Copyright # Taylor & Francis Group, LLC ISSN: 1056-4934 print/1 print/1944-708 944-7086 6 online DOI:   10.1080/10564934.2015.1098265 DOI:

ARTICLES

Social Inequalities in Early School Leaving: The Role of  Educational Institutions and the Socioeconomic Context Jeroen Lavrijsen and Ides Nicaise Katholieke Universiteit Leuven

Reducing the number of early school leavers, those who quit education without at least a high school degree, is a key objective of educational policy throughout Europe. Previous research has shown that in particular particular youngsters youngsters from disadvantaged disadvantaged families families face relativ relatively ely high risks of school dropout. In this paper we use data from the 2009 ad hoc module of the Labour Force Survey to examin exa minee how macromacro-lev level el determ determina inants nts inf influe luence nce sch school ool dro dropout pout ris risks ks amo among ng dif differ ferent ent soc social ial groups. Our results indicate that both the design of the educational system (tracking age, extent of vocational education) and characteristics of the socioeconomic context (poverty rate, unemployment patterns) have an impact on the social distribution of school dropout risk.

In the Europe 2020 growth strategy, the European Commission (2010a ( 2010a)) identified a reduction in the number of early school leavers as one of the five headline policy targets. An upper secondary certificate is regarded to be the minimal qualification needed to fully participate in modern society; for example, early school dropout dramatically increases the risk of unemployment, poverty, and social exclusion (Solga, 2002 (Solga,  2002). ). In this article, we consider the issue of early school leaving from an intergenerational perspective. Previously, it has been shown that low educational attainment is often transmitted across generations. In particular, having low-educated parents drastically increases the odds of becoming a school dropout in every single European country (D ’Addio,  2007).  2007). However, this relationship is not equally strong in different contexts; instead, its strength varies considerably across countries. In this paper, we will examine which country-level educational policies and socioeconomic characteristics determine the strength of the link between parental background and the school dropout risk, using the data from the 2009 module of the Labour Force Survey.

Address correspondence to Jeroen Lavrijsen, Research Institute for Work and Society, Katholieke Universiteit Leuven, Parkstraat 47, Bus 5300, B-3000 Leuven, Belgium. E-mail:   jeroen.lavrijsen@kul [email protected] euven.be

 

296

  LAVRIJSEN AND NICAISE

THEORETICAL BACKGROUND A Double Perspective on Early School Leaving In the litera literatur ture, e, two main main perspe perspecti ctives ves on the occurr occurrenc encee of ear early ly sch school ool lea leavin ving g hav havee been been develo dev eloped ped.. First, First, the educat education ional al lif lifee course course perspe perspecti ctive ve (La (Lamb, mb, Markus Markussen sen,, Teese, Teese, Polese Polesel, l, & Sandberg, 2010) has conceptualized early school leaving as the endpoint of a problematic school career. This perspective views school dropout not as a single event, but rather as the result of a long history of poor academic achievement and disengagement from school. Indeed, longitudinal research has identified low achievement as the most important early predictor of school dropout (Alexander, Battin-Pe -Pears arson on et al., al., 2000 2000). ). Mo More reov over er,, stud studen ents ts wh who o ar aree more more enga engage ged d in Entwis Ent wisle, le, & Kabban Kabbani, i, 2001; 2001; Battin school activities have been shown to face significantly lower risks of dropout (Finn,  1989;  1989; Fredricks, Blumenfeld, & Paris, 2004 Paris,  2004;; Lamote, Speybroeck, Van Den Noortgate, & Van Damme,  2013  2013). ). Comple Com plemen mentar tary y to the lif lifee course course perspe perspecti ctive, ve, the rat ration ional al choice choice perspe perspecti ctive ve (see (see Becker Becker,, 1975; 1975; Breen & Goldthorpe, 1997 Goldthorpe, 1997)) is more focused on the decision to leave school. From this perspective, th thee de deci cisi sion on to drop drop ou outt is re rega gard rded ed as a ra rati tion onal al evalu evaluat atio ion n of the the co cost stss an and d bene benefi fits ts asso associ ciat ated ed wi with th staying in school. At the cost side, direct costs of staying in school (e.g., enrollment fees, costs of  text books, transportation costs) are often dominated by indirect costs, in particular, opportunity costs: as students in school often do not earn money, staying in school corresponds to lost income. On the benefits side, obtaining a qualification may lead to better employment chances and higher earnings later in life. Thus, according to this perspective, the balance between the costs and the anticipated benefits determines whether students stay on at school until graduation.

Social Origin Effects on Educational Attainment In this paper, we will focus on how dropout risks are distributed across different social groups. Overall, social inequalities in educational attainment are a well-documented phenomenon. As the landmark volume of Shavit and Blossfeld (1993 ( 1993)) has demonstrated, parental background has not ceased to influence educational outcomes in the industrialized world throughout the 20th century. On the other hand, the association has been shown to vary over time and among different countries (Breen & Jonsson,   2005). 2005). At least to some extent, the right mix of educational and social policies (Ross,  2009)  2009) seems to reduce educational inequalities. The effects of social origin on educational attainment have often been broken down in socalled primary and secondary effects (Boudon, 1974 (Boudon,  1974). ). Primary effects refer to the observation that children from different social backgrounds perform differently in school (Van de Werfhorst & Mijs, 2010 Mijs,  2010). ). This observation has been explained in a multitude of ways, including cultural disadvant adv antage agess (e.g., (e.g., lowerlower-edu educa cated ted parent parentss are less less able able to help help the their ir childr children en wit with h their their school school assign ass ignmen ments ts)) and econom economic ic constra constraints ints (e. (e.g., g., disadv disadvant antage aged d fam famili ilies es have have few fewer er materi material al resources needed to perform well at school, such as study rooms or ICT equipment). Complementary to the primary effects, a secondary effect of social origin refers to the observation that even student stu dentss who are at the same same level level of academ academic ic perfor performan mance ce appare apparently ntly make diffe differen rentt educational decisions depending on their social background (Breen, Luijkx, Müller, & Pollak,  2009).  2009). Both the primary and the secondary effects of social origin are expected to influence the social distribution of dropout risks. The primary effect on achievement leads children from

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  297

different social classes through different educational life courses, culminating in different risks of dropout. The secondary effect leads children and parents from different social groups to different evaluations of the expected costs and benefits associated with staying on at school. For example, examp le, urgent urgent financial financial constrain constraints ts can make disadvant disadvantaged aged families families perceive perceive the balance balance between short-term costs and long-term benefits differently than more advantaged families.

The Importance of the Macro-Context In the literature on early school leaving, most of the attention has been directed toward identifying the individual risk factors predicting early dropout. This approach has generated valuable insights on possible school-level interventions to reduce dropout among students. However, as a recent literature litera ture review on the effectiven effectiveness ess of interventi interventions ons against against dropout dropout concl concludes udes (Freeman &  2014,, p. 44),   “there is a surprising lack of emphasis in the intervention literature on Simonsen, 2014 Simonsen, develo dev eloping ping interve interventio ntions ns that that addres addresss lar large gerr commun community ity cha charac racter teris istic tics, s, such such as pov povert erty y…. Intervention research must go beyond the typical school boundaries to mediate these factors. ” In this this paper, paper, we will will approa approach ch early early school school leavin leaving g fro from m a more more sys system temic ic perspe perspecti ctive, ve, focusing on both the structural design of the educational system and on macro-level socioeconomic characteristics. A recent example of the value of such an approach can be found in De ). Detecting several associations between time series on early school-leaving Witte et al. (2013 (2013). rates from different European countries, on the one hand, and an array of educational and socioeconomic characteristics, on the other, De Witte et al. (2013 ( 2013)) provided an argument for more system sys temic ic reform reformss on top of school school-le -level vel interv intervent ention ions. s. In partic particula ular, r, De Witte Witte et al. (2013, 2013, p. 340) demonstrated that   “a favourable socio-economic environment (economic growth, the prevention of youth unemployment, the fight against poverty and effective integration strategies for newly arrived immigrants) contributes to more successful school completion rates. ” De Witte et al. (2013) limited their analysis to the effect of macro-level determinants on the overall dropout rates. However, it seems justifiable to add a focus on how such determinants would wou ld influe influence nce dropou dropoutt ris risks ks among among differ different ent social social groups groups.. As rec recent ently ly arg argued ued by Ros Rosss and Leathwood (2013 (2013), ), early school leaving is often only one aspect of a broader context of  social exclusion and marginalization. Hence, it does not only matter to what extent educational or social characteristics influence the overall dropout rate, but also to what extent they perpetuate social inequalities by reproducing low educational attainment and disadvantage over generations. Note that the empirical setup of De Witte et al. (2013 (2013)) relied on yearly data from the core questionnaire of the Labour Force Survey, which do not contain information on parental background. To investigate social inequalities, we will turn below to a specific module added to the Labour Force Survey in the 2009 wave, which does contain such information (see below).

MACRO-LEVEL DETERMINANTS: LITERATURE REVIEW Educational Macro-Level Determinants From the literature, two sets of structural determinants can be distinguished: a set of educational practices, and a set of socioeconomic policies and circumstances (Kritikos & Ching,   2005). 2005).

 

298

  LAVRIJSEN AND NICAISE

Beginning with the former, as dropout often occurs as the endpoint of an educational career marked mar ked by failur failure, e, it is partic particula ularly rly releva relevant nt to consid consider er how the design design of the educat education ional al system deals with disadvantaged or low-achieving students (Lamb et al.,   2010). 2010). A common response in many nations has been to separate low achievers from their academically more talented peers by placing them into less demanding tracks at an early stage in their career (10 –14 years). It has been repeatedly shown that such early tracking increases the gap between strong and weak students (Hanushek & Woessmann, 2006 Woessmann,  2006)) and between students from different social Suchaut,  2005;; Horn, 2009 Horn,  2009;; Van de Werfhorst classes (Dupriez & Dumay, 2006 Dumay,  2006;; Duru-Bellat & Suchaut, 2005 & Mijs, 2010 Mijs,  2010), ), as it raises a number of structural barriers for disadvantaged students to proceed successfully (Husén, 1975 (Husén,  1975). ). Hence, we expect that early tracking will also reinforce the effect of  social background on the probability of graduation (Brunello & Checchi, 2007 Checchi,  2007;; Pfeffer, 2008 Pfeffer,  2008). ). At the same time, the vocational tracks that cater to less academically inclined students may act as a safety net against dropout. In particular, when vocational education delivers specific skills highly demanded by the labor market it may provide access to relatively safe, well-paid jobs. In such cases, vocational education can act as an incentive to stay in school (Shavit & Muller, 2000; Teese, 2011 2000; Teese,  2011). ). Note that a well-developed vocational education track seems particularly rel), which can be combined with evant in the upper secondary cycle (Bol & Van de Werfhorst, 2013 Werfhorst,  2013), a comprehensive structure in lower secondary (such as in the Scandinavian systems). Third, it has been suggested that grade retention often amplifies problems of achievement and motivation instead of solving them (Jimerson, Erson, & Whipple, 2002 Whipple,  2002). ). Students who have  1994;; Stearns, Moller, Blau, & repeated grades are thus more likely to drop out early (Roderick,  1994 Potochnick, 2007 Potochnick,  2007). ). Finally, raising the legal school leaving age could work as a barrier against premature dropout (Cabus & De Witte, 2011 Witte,  2011). ). However, the international variation in the legal school-leaving age is too low to provide convincing statistical checks in a comparative perspective.

Socio-Economic Determinants The second set of determinants refers to the socioeconomic context in which dropout decisions are made. First, the opportunity cost of staying on at school —and thus foreg foregoing oing pote potential ntial —

income depends on how easily young school leavers can find a job. Hence, a high youth unemployment rate is expected to decrease the dropout rates (see, e.g., Petrongolo and San Segundo (2002 (2002)) for Spain or Clark (2011 (2011)) for the UK). However, whether high unemployment may have differential effects on dropout rates among different social groups is still a matter of  debate. On the one hand, disadvantaged youth seem to be more sensitive to the prevailing labor market conditions, as the opportunity costs of staying in school are more heavily felt by those in financial need (Tumino & Taylor, 2013 Taylor,  2013). ). On the other hand, disadvantaged children seem particularly sensitive to the damaging effect of economic crises on the educational aspirations of  children (Rampino & Taylor,  2012).  2012). To the best of our knowledge, social differences in the impact imp act of youth youth unempl unemploym oyment ent on school school dropou dropoutt have have not yet been been ana analyz lyzed ed in an interinternational comparative study. Another obvious socioeconomic factor determining the risk of early school leaving is the poverty rate (Ensminger, Lamkin, & Jacobson,  1996).  1996). First, widespread poverty may influence the primary effect of social origin on performance, for example, through stronger material deprivation

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  299

(e.g., more parents not being able to afford a computer, school books, a study room). Second, povert pov erty y may influe influence nce the percep perceptio tion n of (oppor (opportun tunity ity)) costs: costs: when when pov pover erty ty is more more sev severe ere,, additional income is more badly needed, and disadvantaged families may feel more inclined to give up long-term educational aspirations. It has also been argued that with a more acute threat of poverty, better-off parents will mobilize more of their resources to ensure that their children stay ahead (European Group of Research on Equity of the Educational Systems,  2005).  2005). A final determinant is the expected benefit that is associated with obtaining a high school qualification: do the advantages that the qualification will yield in the future outweigh the costs to be made to obtain it? In the literature several proxies have been employed to quantify the benefits of obtaining a qualification. Some scholars have used the adult unemployment rate, arguing that when adult jobs are scarce, the qualifications paving the way to them would also lose some of their value (“discouraged student effect,” Tumino and Taylor, 2013 Taylor,  2013). ). However, the validity of  this proxy could be questioned, as it assumes that rising adult unemployment would affect everyone to the same degree, independent of the attained educational level. Obviously, this is not the case, as particularly better-qualified groups are less threated by unemployment (Solga, 2002). In this sense, it could even be argued that rising adult unemployment would act as an 2002). incentive, not a discouragement, to obtain a qualification. Other approaches have used GDP/  capita or GDP growth as proxies for the expected future benefits, arguing that economic development would raise the demand for skilled labor and thus increase the labor market value of  high school degrees (Cabus & De Witte, 2012 Witte,  2012,,  2013  2013). ). However, a macro-economic figure such as GDP seems to be a somewhat crude indicator of differences in the value of qualifications across countries. For example, this proxy does not take into account that the value also depends on the design of the labor market, as countries differ in the extent employers rely on educational credentials when hiring employees (Gangl, 2001 (Gangl,  2001;; Maurice, Sellier, & Silvestre,  1986).  1986). In this paper, we will capture the expected benefits of qualifications in the labor market more straightforwardly by calculating the difference that a high school qualification makes on the labor marker for each country; that is, we will use the odds ratio of having a job for adults with and without a high school qualification as the proxy for the expected benefits of this qualification.

MODEL SPECIFICATION In order to examine the effects of educational and socioeconomic factors on the dropout probabilities of individuals from different social groups, we will estimate a multilevel logit model. Multilevel models allow for the interaction between individual characteristics (parental background) groun d) and macro-lev macro-level el variables variables,, while while accommod accommodating ating for the clus clusterin tering g of individua individuals ls within countries through the estimation of country-level errors (Gelman & Hill,   2007 2007). ). Our multilevel model takes the general form











Y ijij  ¼  a  þ  b:S  j  þ  " j  þ   a þ b :S  j  þ  " j  PBij  þ  c: X ijij

 

ð1Þ

in which Yij  denotes the logit of the probability of school dropout of individual   i  in country   j, PBij   indicates parental background, S j  refers to (sets of) educational and socioeconomic characteristics, Xij  is a vector of individual controls and   " j   and   "’ j  are normally distributed random errors at the country level. Hence, if  b  b  is positive, then a higher value of the macro-variable S j  is

 

300

  LAVRIJSEN AND NICAISE

associated with a higher dropout probability. Moreover, we define PB ij ¼ 0 for students who do not have at least one parent with a high school degree, PB ij ¼ 1 for those with at least one parent who finished high school, and PBij ¼ 2 when at least one parent has a tertiary degree. Therefore, a negative estimate for the effect of PBij  implies that dropout probabilities decrease when parents are higher educated, and more negative estimates correspond to stronger effects of parental background. Hence, a positive estimate for  b means that a higher value for S j  is associated with a smaller effect of parental background. Finally, note that we are working with a cross-sectional dataset with only limited time variability (see below) and, consequently, we cannot interpret any of these associations as causal effects. ′

DATA DEFINITION Micro-Level Data for all micro-level variables (i.e., Y ij, PBij and Xij) are drawn from the 2009 ad hoc module modul e of the Labour Force Survey on Entry of Young People into the Labour Market (Eurostat, 2012). 1 In line with the definition of De Witte et al. (2013 ( 2013), ), we define an early school leaver (ESL) as a person who has left   t2he formal education system without having acquired a qualification of at least ISCED level 3. We include respondents aged 20–30 and use gender, age, and having a foreign country of birth as individual controls. To account for different effects of immigration across countries, we allow the effect of foreign origin to vary randomly across nations.

Macro-Level The literature review suggested collecting macro-level data on three educational characteristics (tracking age, extent of vocational education in upper secondary, and incidence of grade retention), and three socioeconomic determinants (youth unemployment rate, poverty rate, benefit of  a high school qualification in terms of the odds on employment). As in the Labour Force Survey age data are aggregated into five-year bands, we take the average of each macro-variable over 5 year bands; we assume that dropout decisions have been made when respondents were about 17 1

Another recent dataset which contains data on parental background and qualification level is PIAAC, a survey organized by the OECD in 22 countries (both from inside and outside Europe; OECD, 2013). Moreover, this dataset contains measures on the skill level of respondents. However, we prefer to use the Labour Force Survey module because this has information on  all  European Union member states and because sample sizes for the corresponding cohorts are on average about five times larger. 2 Note that this definition differs from the official definition of Early School Leaving (EU-ESL) by the European Commission (European Commission, 2010b Commission,  2010b), ), as explained by De Witte et al. (2013 ( 2013). ). In fact, EU-ESL refers to a subset of ESL as we define it here; the official definition counts only those ESL who did not participate in any kind of  nonformal training in  the 4 weeks before the survey. Nonformal training refers here to all courses, seminars, conferences, private lessons, or instructions outside the regular education system, both job-related and for personal purposes. There are two problems with this official definition (see De Witte et al.,  2013  2013). ). First, the primary policy target is lack of  qualifications, not the occasional participation in non-formal training, which is so broadly defined that it includes also small courses for personal purposes that are far less important in terms of long-term consequences. Second, because nonformal training refers to such a vaguely defined spectrum, this component is likely to generate extra noise in the figures.

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  301

TABLE 1 Descriptive Statistics

Country

Sample size

 Early school le leav aver erss ra rate te

AT BE BG CH CZ DE DK EE ES

3,928 3,030 3,466 4,705 5,390 4,512 1,840 1,362 11,842

9% 15% 17% 9% 6% 13% 18% 14% 34%

                 

Co Coun untr tryy

Sample size

 Early school leav leaver erss ra rate te

FI FR GR HU IE IS IT LT LU

3,481 5,602 8,178 8,891 9,675 633 1,5276 1,783 1,878

9% 15% 21% 12% 16% 23% 24% 11% 13%

                 

Coun Countr tryy

Sample size

 Early school leavers rate

LV NL NO PL PT RO SE SI SK

1,248 12,262 2,549 6,748 4,460 6,195 6,495 2,348 3,401

17% 16% 18% 6% 40% 20% 9% 6% 4%

Source: LFS ad hoc module, 2009; 20–30 year-olds.  Note: AT: Austria, BE: Belgium, BG: Bulgaria, CH: Switzerland, CZ: Czech Republic, DE: Germany, DK: Denmark, EE: Estonia, ES: Spain, FI: Finland, FR: France, GR: Greece, HU: Hungary, IE: Ireland, IS: Iceland, IT: Italy, LT: Lithuania, LU: Luxembourg, LV: Latvia, NL: Netherlands, NO: Norway, PL: Poland, PT: Portugal, RO: Romania, SE: Sweden, SI: Slovenia, SK: Slovakia.

years old. For example, we use the averages over the period 2002 –06 as characteristic for the dropout context of the cohort aged 20 –24 (in 2009). Finally, we standardize all macro-variables to have an average of 0 and a standard deviation of 1 across all countries. Data on tracking age and on the incidence of grade retention among 15-year-olds are derived from PISA (2003 and 2009 waves), while data on the percentage of upper secondary students in vocational education are derived from Eurostat. All data on the socioeconomic macro-level determinants are derived from Eurostat. We define poverty in terms of the absolute material deprivation rate (not having adequate material resources). Results using the relative poverty rate (earning less than 60% of the median income) are equivalent, but weaker. Finally, the indicator reflecting the benefit ben efit of qualifi qualificat cation ionss is define defined d as the odds odds ratio ratio of une unempl mploym oyment ent bet betwee ween n adu adults lts withou withoutt and with a secondary qualification; the higher the value, the more difference a qualification makes.

Sample The LFS ad hoc module contains data on 30 European countries. As we did not find sufficient information on the macro-level variables for Malta, Cyprus, and the UK, these countries were removed from the sample. The sample sizes and the rates of early school leavers in the 27 remaining countries are presented in Table 1 Table  1.. The full sample consists of 141,178 respondents.

RESULTS Model Without Macro-Level Determinants We now turn to examining which country-level determinants explain the strength of the effect of parental background on dropout risks. We start with a basic model that predicts early school

 

     3      1

    *     *      5     7     1  .  .     1     0     0

    *     *     *     *     *     *     *     *     *     2     6     0     3     8     7  .     3  .     4  .     7  .     0  .     0     0     0     0     1

    *     1     6     0  .  .     1     0     0

    *     *     7     7     0  .  .     2     0     0

    *     *     *     *     *     *     *     *     *     2      5     2     4     6     0     7  .     4  .     3  .  .     6  .     0     0     0     0     1

    *     *     *     *     0     9     2  .     1  .     0     0

    *     *      5     1  .     0

    *     *     *     *     *     *     *     *     *     3      5     4     3     4     0     7     4  .  .  .     6  .     3  .     0     0     0     0     1

    *     *     *     1     8     2  .  .     1     0     0

           

   s     t    n    a    n      i    m    r      2    e      1     t    e      d    c      i    m    o    o    n      1    c    e    o      i    c     s      t    o    s     n     a      0      þ     n      i      l      1     m    a     r    n     e    o      t      i     t     e    a      D    c      l    u     e      d      9     v      E

    e      L        o     r     c     a      M      d     n     a     s     c      i      t     s      i     r     e      t     c     a     r      2     a      h      E      L     C      B     l      A    e     v      T    e      L        o     r     c      i      M      f     o     s     m     r     e      T     n      i     g     n      i     v     a     e      L      l     o     o

    *     *     *     *     *     *     7      5     2  .     2  .     0     0

    *     *     *     *     *     *     *     *     *     1     6     4     4     6     7  .     4  .     3  .     7  .     0  .     0     0     0     0     1

       

           

           

   

   

   

   

       

       

       

   

    *     *     *     6     2  .     0

    1     0  .     0

     8

    *     *     *     *     *     *     *     *     *     4      5     2     1     6     4     7     0  .  .  .     6  .     3  .     0     0     0     0     1

    *     *     6     8     0  .     1  .     0     0

     7

    *     *     *     *     *     *     *     *     *     1     6      5      5     7     4     0     7  .  .  .     7  .     3  .     0     0     0     0     1

    *     *     *     *      5     7     8     0  .     2  .     1  .     0     0     0

    *     *     *     *     *     *     2      5     6     2     4     7  .     4  .     0  .     3  .     7  .     0     0     0     0     1

    *     *      5     2  .     0

   s     t    n    a      i    n    m    r    e     t    e      6      d    c      i    m    o    n    o    c      5    e    o      i    c    o      S      4

   s     t    n    a      3    n      i    m    r    e     t    e      d      h     c      l      2      S    a    n    o     y      i      l     t     r    a     a      E    c    u      d      E      1

       

           

           

   

   

       

   

   

   

   

     5     0  .     0

       

    *     *     *      5      5  .     0

   

   

    *     *     *     2     3  .     0

   

    4     1  .     0

     1

     1      7      1  ,      2      2      0      1

     4      7      2  ,      2      2      0      1

    *     *     *     3     6  .     0

   

     6      7      3  ,      2      2      0      1

   

   

    2     0  .     0

    *     *     6     2  .     0

    *     *     *     4     4  .     0

    *     3     1  .     0

    *     *     1     3  .     0

     6      7      1  ,      2      2      0

   

    *     *     6     1  .     0

    3     4     2     0  .     1  .     0  .     0     0     0

    *      5     1  .     0

    *     *     *     0     6  .     0

    1     0  .     0

   s      7      0      7      2      2  ,      0      1      5      7      0  ,      2      2      0      1

   

   

    *     *     1     3  .     0

   

    *     4     1  .     0

    *     *     *     6     4  .     0

   

   

    *     *     *     *     *     *     *     *     *     4      5     1     4     4     7  .     4  .     0  .     3  .  .     6     0     0     0     0     1            

    *     *     *     0     4  .     0

   

   

    *     *     *     *     *     *     *     *     *     3      5     1     8     6     7  .     0  .     3  .     6  .     4  .     0     0     0     0     1            

    *     7     1  .     0

    *     *     8     1  .     0

    *     *     *     *     *     *     *     *     *     2     6     2     3     6     7  .     0  .     3  .     7  .     4  .     0     0     0     0     1            

    *     *     *     0     4  .     0

    *     *     *     3     4  .     0

    1      5     0  .  .     1     0     0

   

    1     0  .     0

    *     *     *     1     3  .     0     *     *     *     1     3  .     0

   

    *     *     *     3     2  .     0

    3     0  .     0

    *     *     *     *     *     *     *     *     *     2     6     1     2     6     7     0     4  .  .  .     7  .     3  .     0     0     0     0     1            

   

    *     *     *      5     4  .     0

   

    *     *     *     7      5  .     0

    1     0  .     0

    *     *     *     7     2  .     0

   

       

    *     *     *      5     2  .     0

    3     0  .     0

     5      7      1  ,      2      2      0      1

    2     1  .     0

     3      7      4  ,      2      2      0      1

   

     5      7      1  ,      2      2      0      1

     1      7      4  ,      2      2      0      1

     4      7      4  ,      2      2      0      1

    3     0    –     0     2    ;     9     0     0     2  ,    e     l    u     d    o    m    c    o     h     d    a     S     F     L    :

   e    c    r    u    o      S

 .  .    g     1  .    n     0     *     *     *     i     *     *     *      2    v      <     *     *     *     *    a     3     4     4      5     1     4     4     7     2      7      4      2      0      2  ,    p     1  .     4  .     0  .     3  .     6  .     0  .     7  .     2  .     0  .     l    e     *     0     0     0     0     0     0     1     0     0      1     l  ,    o     5                                o     0     h  .    c     *     *     *    s     0     *     *     *      6     *     *     *      4    y      <     4      5     2     1     4      7      2     l    p  ,     7    r      2      0  .     0  .     4  .  .     3  .     7    a     *     d     0     0     0     0     1      1     *    e    n                    :  ,    u    e     1    o     l    r     0  .    g     b    s     h    s     t     k    a     0    n    n     i    u    c    e    o    o    r     t    e     i    a    o     i    e     t    a    a      <     t     b     t    r     t    a     Y    a     T    a    v    p     T    a     l    c    r     t    c    r     i     E     *     *     *     *     *    a     *     i     *     E     t    n     f     t     f     d     d     d     d     d     d    n     d     i    n     i    n     *     V    e    o     l    n    n    n    n     V    n    n     l    e    n     i    n     t    s    o    e     *    a    e     i     f    a    m    t    u    u    u     f    u    c    u    a    u     t    u    u    r    u     i     d    :    o    y    a    q    o    o    o    o    o    s    a    o    o     t    r    o    a     l    r    r    r    r    r    r    n    r    q    s    n     t    o    v    p    v    g     f     i    o     i    g    g    g     t    g    n    g     i    e    e    g     t    c    o     l    n     i    n     k    r    o     k    r     f    r    n     k    p    p     i    r     k    o     k    e     k    n     k     i     h     t    p    e    o    e    n    e    e     t     t    e     t     t    p    g    c    c    c    c    c    c     t     t    c    e    g    n     i    a    g    n    m     i    e      C    x    e    e     d    e    a     t    a    a    a    x    a    n    c    a    m    a    e    a    e    s    a    e    r     b    o      I    c    e    c    w    a     d     b     b     b     b     b    e     D     t     b    c     i     t    o    c     t    y    n    r      A    n    n     f    e     l    g    n    g    e     l     l     l     l     l     l     l     l      e    o    a    :    u     l    p    e    n     i    r    e    a    a    r    a     t    n    v    n     t    a    a     t    a    n     i    a     i    a    v     t    a    a     i    a     l     t    a     t    n     t    o     i    a    e     l     h    e    o     t    g    n     i    r    r     M    i    e     h      d     t    n    u    c    e     t     t     d     M   c     k     i     i    a     t    n     k    n     i    n    r    n    p    n    r     t     t     t    n    g    e    o     L    r    e    o    o     i    u    a    e     d     d    c    m    e    c    e    e    e    e    e    e     l    c    a     t     L    u    m     l     i     t    e    m    e    a    o    a    p    r    r    r    a    e    r     l    r    a    r    r    r    e    r    a    r    p    e    a    e    a    g     S      N    r    r    v    s    e    a    o    r    a    a    e    a    a    a    a    r    e    n    n    n     d    m     i    n    m

   e     P     T     P     R     P     G    P    u     P     M    P     I    o    o      N     P     d     I     F     A    F     P     t    e     T     R     G    Y    M    I    o     t    n    n     I     I     D     M 302

 .    s     d     l    o    r    a    e    y

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  303

leaving on the basis of individual characteristics, but without including any macro-level predictors. The results are presented in Model 1 in Table 2 Table  2.. The estimates indicate that being female reduces the odds of school dropout, while being of foreign origin increases them. Importantly, the negative negative estimate of parental parental background background indicates indicates that having having higher higher educa educated ted parents parents reduce the probability to become a school dropout. On average, the odds to become a dropout are multiplied by a factor of about 0.25 for each step upward in the parental background scale; the odds on dropout are thus divided by a factor 4 when shifting from low- to middle-educated parents or from low- to high-educated parents. Howeve How ever, r, this this averag averagee value value masks masks large large differ differenc ences es betwee between n cou countr ntries ies.. The var varian iance ce components of the model (i.e., the estimated variance of the random terms   " j  and   "’ j) are markedly larger than their standard errors (for the intercept: 0.71, with a SE of 0.11; for the slope of  parental background: 0.55; with a SE of 0.09). This indicates that there is relevant cross-country variability, both in the absolute dropout rates and in the social origin effects and that a multilevel model is appropriate. To present this more illustratively, the estimated residual terms for the slope (the country-specific deviation from the average value of     1.34) were converted into odds ratios and plotted as crosses in Figure  1.  1 . This figure shows that the average effect on the odds ratio of dropout of each step upward in the parental background scale (0.25 on average) can be broken down in country-specific odds ratios ranging from 0.05 in Bulgaria to 0.69 in Denmark. Put otherwise, shifting from having low- to medium-educated parents reduces the dropout odds by a factor 20 in Bulgaria, as against a factor 1.5 in Denmark. In particular, the Nordic countries (Denmark, Finland, and Iceland) show relatively low levels of inequality in risks, while a number of Southern and Central European countries record high inequality levels.

FIGURE 1  Country-specific deviations from the estimated average odds ratio on dropout for each step upward in the parental background scale, both before including any macro-level determinants (Model 1) and after including a number of macro-determinants (Model 13). Odds ratios closer to 1 indicate smaller effects of parental background.  Source: LFS ad hoc module, 2009; 20–30 year-olds.

 

304

  LAVRIJSEN AND NICAISE

The other twelve models in Table   2  include the different macro-level determinants (tracking age, the relative share of vocational education, grade retention, youth unemployment rate, material deprivation rate, the importance of qualifications in the labor market). The results are presented in three thr ee blocks blocks of rows. rows. The fir first st rows rows lis listt the effect effect of indivi individua duall cha charac racter terist istics ics (gende (gender, r, age, age, foreign origin, parental background) on the dropout probability. In all models the estimates are very similar to Model (1). A second block of rows represents the  main effect  of   of the macro-level determinants, that is,  b  in specification (1). The third block of rows represents the impact of the macro-level determinants on the effect of parental background, that is,   b in specification (1). Recall that as the dependent variable is the logit of the dropout probability, a positive estimate for   b  means that an increase in the value of the macro-variable is associated with an increase in the odds of dropout, while a positive estimate for the interaction term   b means that a higher value of the macro-variable is associated with a smaller effect of parental background on dropout. ′



Educational Determinants The first important finding of the analysis is that a later tracking age seems to be associated with a lower effect of parental background on early school leaving. This effect is consistent across all spec specifica ification tions, s, with significa nt effect effe ct tracking sizes. sizes. This findingthe confirms confi rmsofthe expectati expectation on from the literature, which hassignificant shown that early increases effect parental background on academic performance (Dupriez & Dumay,   2006; 2006; Duru-Bellat & Suchaut,  2005  2005;; Horn,  Horn,   2009; 2009; 2007; Van de Werfhorst & Mijs,  2010)  2010) and on educational attainment (Brunello & Checchi,   2007; Pfeffer, 2008 Pfeffer,  2008). ). The effect of tracking age on the absolute dropout probability is small; it reaches significance in only two models. Second, the extent of vocational education (in upper secondary) is consistently negatively associated with the overall dropout rate; the estimates are significant in most specifications. This confirms the expectation that vocational education acts as a safety net, encouraging young people to stay on at school because of the attractive labor market returns of vocational degrees (Shavit (Sha vit & Muller, Muller,   2000), 2000), alt althou hough gh a revers reversee interp interpret retati ation on (an exp expans ansion ion of the vocati vocationa onall sector to absorb a growing number of youngsters staying longer in education) can also be part of the explanation. The safety net of vocational education seems to work for everyone to the same degree, as there is no clear interaction with parental background. Finally, the grade retention rate has no clear effects, neither on the overall dropout rate nor on the effect of parental background. Note that earlier findings on the micro-level consistently showed that grade retention is an important predictor of  individual  individual  dropout (Jimerson, Erson, & Whipple, 2002 Whipple,  2002;; Roderick, 1994 Roderick,  1994;; Stearns, Moller, Blau, & Potochnick, 2007 Potochnick,  2007). ). Apparently, this does not translate in any association between the aggregated grade retention rate and the aggregated dropout rate at the country level. A possible explanation is that in countries that make less extensive use of grade retention, problems of underachievement and disengagement are not always adequately dealt with neither.

Socio-Economic Determinants Concerning the effect of the socioeconomic context, we would first expect high youth unemployment to be associated with a lower dropout rate, as the pull effect of the labor market

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  305

Rumberger,  1983). ). However, the models decreases when jobs are sparser (De Witte et al.,  2013;  2013; Rumberger, 1983 in Table  Table   2  show that a high youth unemployment rate seems not to be associated with a lower dropout probability among disadvantaged students (reference group). However, there is a strong intera int eracti ction on effect effect betwee between n youth youth unempl unemploym oyment ent and parent parental al backgr backgroun ound, d, whi which ch means means thatt youth tha youth unempl unemploym oyment ent does does have have the expect expected ed effect effect for respon responden dents ts with with mid middle dle-- and high-educated parents. These tendencies seem to support the argument developed by Rampino and Taylor (2012 (2012)) which states that in times of economic crisis, mass youth unemployment particularly seems to damage the educational aspirations of disadvantaged students, and these reduced educational aspirations seem to outweigh the effects of lower job availability. The educational aspirations of students with high-educated parents are less sensitive to short-term economical fluctuations, and for this group the availability of jobs (decreased pull) seems more important. However, note that the reverse interpretation cannot be excluded; that is, the youth unemployment rate may also react to changes in the inflow of unqualified school leavers on the labor market. From Fro m all socioe socioecon conomi omicc determ determina inants nts,, the effect effectss of povert poverty y rat ratee are the str strong ongest est.. The material mater ial deprivati deprivation on rate—the proportio proportion n of people people living living in mate materiall rially y inadequa inadequate te circumcircumstances—seems intensely associated with a high dropout rate among disadvantaged respondents and with a strong effect of parental background. In all model specifications, the estimates are significantly negative and the (standardized) estimates are larger than that of the any other variable. The strength of the effect of poverty on dropout among disadvantaged students may be explained by the fact that poverty determines their trajectory in a multitude of ways. First, when poverty is severe, disadvantaged families may lack even the essential resources to give children a fair chance at school (e.g., a study room), which may increase the gap with better-off students. At the same time, poorer families badly need the additional income that could be earned by a school leaver, and hence the pull of the labor market becomes stronger. Note that even when we would not expect strong effects of poverty among high-educated families (since these are often situated near the top of the income distribution), it is interesting to see that a high poverty rate ap appa pare rent ntly ly is asso associ ciat ated ed with with a   lower   dropo dropout ut amon among g chil childr dren en of we well ll-o -off ff pa pare rent ntss (t (the he interaction effect is much larger than the main effect). This seems to confirm the suggestion that in times of economic harshness well-off parents invest more of their resources to secure a strong ed educ ucat atio iona nall posi positi tion on fo forr th thei eirr ch chil ildr dren en (E (Eur urop opea ean n Grou Group p of Re Rese sear arch ch on Eq Equi uity ty of the the Educational Systems, 2005 Systems,  2005). ). Finally, the main effect of our indicator for the benefits of a qualification is always significantly negati neg ative. ve. This This means means that that when when high high school school qualif qualifica icatio tions ns ma matte tterr more more in the labor labor marke markett (qualifications lead to better job chances), students seem more inclined to stay at school. The interaction effect with parental background is less clear, with some estimates suggesting that the effect is less strong for advantaged students, but the estimates are not always consistent.

Explanatory Power In order order to assess assess how well well the differ different ent macromacro-var variab iables les exp explai lain n the differ differenc ences es betwee between n countries that were observed before, the country-specific deviations from the average slope that persist after  including   including macro-level determinants (i.e., the residuals estimated in Model 13, converted into odds ratios) were added as circles to Figure   1. Apparently, the large discrepancies

 

306

  LAVRIJSEN AND NICAISE

that existed between countries in the parental background effect are drastically reduced when differences in the macro-level determinants included this model (tracking age, relative extent of VET, VET, povert poverty y rate, rate, and import importanc ancee of qualif qualifica icatio tions) ns) are taken taken into into acc accoun ount: t: the circle circless are distributed far more narrowly around the average than the crosses.

EDUCATIONAL OR SOCIOECONOMIC INTERVENTIONS? One of the long-standing issues in educational research concerns the role that education can play in equalizing educational opportunities in an unequal society. For example, in his book  1951,, p. 142) already compellingly claimed  Equality from 1931, the social critic R. H. Tawney ( 1951 that schools could never stand up to such high expectations   “as though opportunities for talent to rise could be equalized in a society where the circumstances surrounding it from birth are themselves unequal!”  More recently, the potential of educational reforms to reduce the strength of the parent parental al backgr backgroun ound d has been been called called into into questi question. on. Instea Instead, d, it has been been increa increasin singly gly argued that it is not educational reforms but rather an equalisation of cultural and economic resources that are the prerequisite for educational equity (Erikson & Jonsson,  1996  1996;; Esping 2004;; Shavit & Blossfeld,  1993  1993). ). Further note that educational and socioeconomic Andersen, 2004 Andersen, determinants determina nts are often correlate correlated d (Allmend (Allmendinger inger & Leibfried Leibfried,,   2003 2003;; Hega Hega & Hokenm Hokenmaie aier, r, 2002;; West & Nikolai, 2013 Nikolai,  2013); ); in particular, educational reforms, like the  “ comprehensivization” 2002 of lower secondary education in Scandinavia, often have been designed as part of a larger welfare policy toward a more egalitarian society (Antikainen,  2006  2006). ). Previo Pre viousl usly, y, the effect effectss of the educat education ional al system system and the eff effect ectss of the soc socioe ioecon conomi omicc environment have been often studied in isolation. For example, when Brunello and Checchi 2007)) studied the effect of the educational tracking policy on social inequality in high school (2007 dropout, they controlled for the confounding effects of other educational characteristics (such as the extent of vocational education), but they did not take into account the socioeconomic context. However, as the analysis in this paper demonstrates, the context does play an important role in understanding the dropout probabilities of disadvantaged students. Our present analysis enables us to disentangle the effects of both educational and the socioeconomic features. First, the observed effects have proven robust against mutual control. For example, both the tracking age and the poverty rate seemed associated with the strength of  the parental background effect, and these effects persisted in those models that contained both determinants (Model 9, 12, and 13). To illustrate the size of the effects of different characteristics in a model with mutual control, Figure  Figure   2  plots the estimated impact of the four macrovariables that were included in Model 13—tracking age, relative extent of VET, poverty rate, and importance of qualifications—on the dropout rates of children from different social groups. For each macro-variable, the estimated dropout probabilities are plotted for two situations: one in which the value of the macro-variable is one standard deviation above the countrywide average, and another where this value was one standard deviation below average (all other things being equal). The figure demonstrates that both the educational and the socioeconomic characteristics have sizeable effects, also after mutual control: tracking and poverty clearly affect the slope of parental background, while the extent of vocational education and the importance of  qualifications influence the absolute level of dropout. At the same time, a clear effect of parental background remains in all situations.

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  307

FIGURE 2   Estimated impact of the four macro-variables from Model 13 on the dropout rates of children from different social groups.  Source: LFS ad hoc module, 2009; 20 –30 year-olds. For each macro-variable, the estimated rate is plotted for the situation where the value is one standard deviation above resp. below the average across all countries.

Considering the above results, adjusting the design of the educational system certainly seems to have an effect on equalization of opportunities, in addition to the effect of broader welfare policies. In this sense, the hypothesis that prior equalization of socioeconomic conditions in society at large is a  conditio sine qua non  to achieve greater equality of educational opportunities (Esping 2004)) seems not entirely true: there is certainly room for improvement within the eduAndersen, 2004 Andersen, cational system as well. On the other hand, there is no doubt that the socioeconomic context plays an important role in equalising educational outcomes (Bryan, 2005 (Bryan,  2005). ). Hence, our findings strongly connect to an observation put forward by T. Husén (1975, p. 164) that   “there is a growing realization that educational reforms must be coordinated with social and economic reforms. Indeed, it is impossible to establish better equality of opportunity in the educational system without its being established previously or simultaneously in the overall prevailing social system. ”

CONCLUSION In this paper we use data from the 2009 ad hoc module of the Labour Force Survey to examine the impact of macro-lev macro-level el socioeco socioeconomic nomic and education educational al determinan determinants ts on the social social distridistribution of early school dropout risks. The results concerning the effect of educational system characteristics are twofold. First, a large vocational sector in upper secondary education functions seems to act as a safety net against dropout by offering less academically inclined students a valuable alternative with relatively attractive labor market prospects. Secondly, while early tracking has no consistent effects in terms of average dropout rates, it is consistently associated with a larger effect of parental background.

 

308

  LAVRIJSEN AND NICAISE

In regard to the influence of the socioeconomic context, we find a strong effect of the poverty rate on social inequalities in early school leaving: children from low-educated parents are far more likely to drop out when poverty is high. High youth unemployment is also associated with high dropout probabilities for disadvantaged students. Finally, we detected an inverse association between dropout rates and the returns on academic qualifications (in terms of reduced unemployment risks) in adult life. The approach taken in this paper considered early school leaving not just as an individual issue, to be prevented by school-level interventions, but rather as a phenomenon that is strongly mediated by a number of macro-level features. Consequently, our results may inform educational policy about how altering the organizational design of the educational system may complement ). At the same time, our findings school-level interventions against dropout (Lee & Burkam, 2003 Burkam,  2003). show that social inequalities in educational attainment are not only a result of the way the educational system functions, but also of socioeconomic inequalities outside the reach of schools. A major drawback of our research design is that we had to use a cross-sectional dataset, which impedes imp edes making making strong strong causal causal claims. claims. For some some of our macro-le macro-level vel dete determin rminants ants,, bidirect bidirectiona ionall mecha me chanis nisms ms are are not not unlik unlikely ely.. For exam example ple,, the corre correlat lation ion betwe between en schoo schooll dropo dropout ut and you youth th unemunemploymentt may be attributed to ploymen to the pull effect of the the labor market market on student students, s, but may also be reflect reflect th thee di diffi fficul cultie tiess to in integ tegrat ratee unqua unqualif lified ied school school leave leavers rs int into o the labor labor mark market. et. A step step forwa forward rd for future future research could be to test the relationships observed in this study on (national) longitudinal datasets.

AUTHOR BIOS Jeroen Lavrijsen is a senior research associate at HIVA –KU Leuven (Research Institute for Work and Society). He investigates the effect of educational system design in the medium-long term (acquisition of qualifications, transition to the labour market) with special attention to patterns of social inequality in these processes. Prof. Dr. Ides Nicaise works as a research manager at HIVA–KU Leuven (Research Institute for Work and Society Society). ). His research research focuses focuses on the economi economics cs of educat education ion as wel welll as on poverty and social exclusion.

REFERENCES Alexander, K., Entwisle, D., & Kabbani, N. (2001). The dropout process in life course perspective: Early risk factors at home and school.  Teachers College Record ,   103, 760–822. doi:10.1111/0161-4681.00134 doi:10.1111/0161-4681.00134.. Allmendinger, J., & Leibfried, S. (2003). Education and the welfare state: The four worlds of competence production.  Journal of European Social Policy,   13, 63–81. doi:10.1177/0958928703013001047 doi:10.1177/0958928703013001047.. Antikainen, A. (2006). In search of the Nordic model in education.  Scandinavian Journal of Educational Research,  50 , 229–243. doi:10.1080/00313830600743258 doi:10.1080/00313830600743258.. Battin-Pearson, S., Newcomb, M. D., Abbott, R. D., Hill, K. G., Catalano, R. F., & Hawkins, J. D. (2000). Predictors of  early high school dropout: A test of five theories.  Journal of Educational Psychology ,  92 (3), 568–590. doi:10.1037/  doi:10.1037/  0022-0663.92.3.568.. 0022-0663.92.3.568 Becker, Becke r, G. (1975 (1975). ).   Human Human cap capita ital: l: A the theore oretic tical al and emp empiri irical cal ana analys lysis, is, wit with h specia speciall ref refere erence nce to edu educat cation ion. New York: Columbia University Press. Bol, T., & Van De Werfhorst, H. (2013). Educational systems and the trade-off between labor market allocation and equality of educational opportunity.  Comparative Education Review,  57 , 285–308. doi:10.1086/669122 doi:10.1086/669122.. Boudon, R. (1974).  Education, opportunity & social inequality: Changing prospects in Western society . New York: Wiley.

 

SOCIAL INEQUALITIES IN EARLY SCHOOL LEAVING

  309

Breen, R., & Goldthorp Breen, Goldthorpe, e, J. (199 (1997). 7). Expla Explaining ining educati educational onal different differentials ials towards a form formal al ratio rational nal actio action n theor theory. y.  Rationality and Society,  9 , 275–305. doi:10.1177/104346397009003002 doi:10.1177/104346397009003002.. Breen, R., & Jonsson, J. (2005). Inequality of opportunity in comparative perspective: Recent research on educational attainment and social mobility.  Annual Review of Sociology ,   31 , 223–243. doi:10.1146/annurev.soc. doi:10.1146/annurev.soc. 31.041304.122232.. 31.041304.122232 Breen, R., Luijkx, R., Müller, R., & Pollak, R. (2009). Nonpersistent inequality in educational attainment: Evidence from eight European countries.   American Journal of Sociology,  114 , 1475–1521. doi:10.1086/595951 doi:10.1086/595951.. Brunello, G., & Checchi, D. (2007). Does school tracking affect equality of opportunity? New international evidence.  Economic Policy,  22 , 781–861. doi:10.1111/j.1468-0327.2007.00189.x doi:10.1111/j.1468-0327.2007.00189.x.. Bryan, J. (2005). Fostering educational resilience and achievement in urban schools through school-family-community partnerships.  Professional School Counseling,  8 , 219–227. Cabus, S. J., & De Witte, K. (2011). Does school time matter? —On the impact of compulsory education age on school dropout.   Economics of Education Review ,  30 , 1384–1398. doi:10.1016/j.econedurev.2011.07.003 doi:10.1016/j.econedurev.2011.07.003.. Cabus, S. J., & De Witte, K. (2012). Naming and shaming in a  ‘ fair’  way. On disentangling the influence of policy in observed outcomes.   Journal of Policy Modeling ,  34 (5), 767–787. doi:10.1016/j.jpolmod.2012.03.001 doi:10.1016/j.jpolmod.2012.03.001.. Cabus, S. J., & De Witte, K. (2013).   Why do students leave education early? Theory and evidence on high school dropout rates  (TIER Working paper series-TIER WP 13/01). Top institute for Evidence Based Education Research, Maastricht, 1–24. Clark, Cla rk, D. (20 (2011) 11).. Do rec recess ession ionss keep keep stu studen dents ts in sch school ool?? The imp impact act of you youth th une unempl mploym oyment ent on enr enrolm olment ent in post-compulsory education in England.  Economica,   78, 523–545. doi:10.1111/j.1468-0335.2009.00824.x doi:10.1111/j.1468-0335.2009.00824.x.. D’Addio, A. (2007).  Intergenerational transmission of disadvantage: Mobility or immobility across generations?  Paris: OECD Publishing. De Witte, K., Nicaise, I., Lavrijsen, J., Van Landeghem, G., Lamote, C., & Van Damme, J. (2013). The impact of  institutio insti tutional nal conte context, xt, educa education tion and labo labour ur marke markett policies policies on early school leaving: A comp comparati arative ve analy analysis sis of  doi:10.1111/ejed.12034.. EU countries.   European Journal of Education,  48 (3), (3),   331–345. doi:10.1111/ejed.12034 Dupriez, V., & Dumay, X. (2006). Inequalities in school systems: Effect of school structure or of society structure? Comparative Education,  42 , 243–260. doi:10.1080/03050060600628074 doi:10.1080/03050060600628074.. Duru-Bellat, M., & Suchaut, B. (2005). Organisation and context, efficiency and equity of educational systems: What PISA tells us.  European Educational Research Journal ,  4 , 181–194. doi:10.2304/eerj.2005.4.3.3 doi:10.2304/eerj.2005.4.3.3.. Ensmin Ens minger ger,, M. E., Lamkin Lamkin,, R. P., & Jacobs Jacobson, on, N. (19 (1996) 96).. Sch School ool lea leavin ving: g: A longit longitudi udinal nal perspe perspecti ctive ve includ including ing neighborhood effects.  Child Development ,  67 (5), (5), 2400–2416. doi:10.2307/1131630 doi:10.2307/1131630.. Erikson, R., & Jonsson, J. (1996).  Can education be equalized? The Swedish case in comparative perspective . Boulder: Westview Press. Esping-Andersen, G. (2004).   Generation Generational al incom incomee mobil mobility ity in North America and Euro Europe pe. Cambr Cambridge idge:: Cambr Cambridge idge University Press. European Commission. (2010a).   Europe 2020 —  A strategy for smart, sustainable and inclusive growth, COM (2010) 2020 final. Brussels. European Commission. (2010b). Reducing early school leaving. Accompanying document to the Proposal for a Council Recommend Reco mmendation ation on polic policies ies to reduce reduce early school leaving. European Group of Research on Equity of the Educational Systems. (2005). Equity in European Educational Systems: A set of indicators. Eurostat. (2012).  LFS Ad Hoc Module 2009 –  Entry of young people into the labour market . Luxembourg. Finn, Fin n, J. (1989) (1989).. Withdr Withdrawi awing ng fro from m sch school ool..   Review Review of Edu Educat cation ional al Res Resear earch ch,   59(2), (2), 117 117–142 142.. doi doi::10.3102/  00346543059002117.. 00346543059002117 Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of the concept, state of the evidence.   Review of Educational Research,  74 , 59–109. doi:10.3102/00346543074001059 doi:10.3102/00346543074001059.. Freeman, J., & Simonsen, B. (2014). Examining the impact of policy and practice interventions on high school dropout and school completion rates: A systematic review of the literature.  Review of Educational Research,  85 , 205–248. 10.3102/0034654314554431.. doi:10.3102/0034654314554431 doi: Gangl, M. (2001). European patterns of labour market entry. A dichotomy of occupationalized vs. non-occupationalized systems?  European Societies,   3, 471–494. doi:10.1080/14616690120112226 doi:10.1080/14616690120112226.. Gelman, A., & Hill, J. (2007).  Data analysis using regression and multilevel/hierarchical models. New York: Cambridge University Press.

 

310

  LAVRIJSEN AND NICAISE

Hanushek, E. A., & Woessmann, L. (2006). Does educational tracking affect performance and inequality? Differences-indifferences evidence across countries.  The Economic Journal,  116 , C63–C76. doi:10.1111/j.1468-0297.2006.01076.x doi:10.1111/j.1468-0297.2006.01076.x.. Hega, G., & Hokenmaier, G. (2002). The welfare state and education: A comparison of social and educational policy in advanced industrial societies.   German Policy Studies,  2 , 143–173. Horn, D. (2009). Age of selection counts: A cross-country analysis of educational institutions.  Educational Research and Evaluation,   15, 343–366. doi:10.1080/13803610903087011 doi:10.1080/13803610903087011.. Husén, T. (1975).   Social influences on educational attainment: Research perspectives on educational equality. Paris: Organisation for Economic Co-operation and Development. Jimerson, S. R., Erson, G. E., & Whipple, A. D. (2002). Winning the battle and losing the war: Examining the relation bet betwee ween n grade grade ret retent ention ion and dro droppi pping ng out of high high sch school ool.. Psy Psycho cholog logyy in the Sch School oolss, 39, 441 441–457 457.. doi: doi:10.1002/pits.10046 Kritikos, E., & Ching, C. (2005).  Study on access to education and training, basic skills and early school leavers  (Ref. DG EAC 38/04). Lot 3: Early School Leavers. Final Report. Lamb, S., Markussen, E., Teese, R., Polesel, J., & Sandberg, N. (2010).  School dropout and completion: International comparative studies in theory and policy. Dordrecht, Netherlands: Springer. Lamote, Lamo te, C., Speybroec Speybroeck, k, S., Van Den Noortgate, Noortgate, W., & Van Damme, J. (2013 (2013). ). Diff Different erent pathway pathwayss towar towards ds dropo dropout: ut: The rol rolee of eng engage agemen mentt in ear early ly school school lea leavin ving. g.   Oxford Oxford Revi Review ew of Educ Education ation,   39(6), (6), 739–760. doi:10.1080/  doi:10.1080/  03054985.2013.854202.. 03054985.2013.854202 Lee, V. E., & Burkam, D. T. (2003). Dropping out of high school: The role of school organization and structure.  American Educational Research Journal,  40 (2), 353–393. doi:10.3102/00028312040002353 doi:10.3102/00028312040002353.. Maurice, M., Sellier, F., & Silvestre, J. J. (1986).  The social foundations of industrial power: A comparison of France and Germany. Cambridge, MA: MIT Press. OECD. (2013).  First results from the Survey of Adult Skills The Survey of Adult Skills . Paris: OECD Publishing. Petrongolo, B., & San Segundo, M. (2002). Staying-on at school at 16: The impact of labor market conditions in Spain. doi:10.1016/s0272-7757(01)00019-x..  Economics of Education Review,  21 , 353–365. doi:10.1016/s0272-7757(01)00019-x Pfeffer, F. (2008). Persistent inequality in educational attainment and its institutional context.  European Sociological  Review,  24 , 543–565. doi:10.1093/esr/jcn026 doi:10.1093/esr/jcn026.. Rampino, T., & Taylor, M. (2012).  Educational aspirations and attitudes over the business cycle  (ISER Working Paper Series: 2012–26). Institute for Social and Economic Research, Essex. Roderick, Rode rick, M. (1994 (1994). ). Grade retentio retention n and schoo schooll dropo dropout: ut: Inves Investigat tigating ing the asso associati ciation. on.   American American Educational doi:10.3102/00028312031004729..  Research Journal,  31 (4), 729–759. doi:10.3102/00028312031004729 Ross, A. (2009).   Educationa Educationall polic policies ies that address soci social al inequ inequality ality,, repor reportt from the EPAS EPASII prog programme ramme. London: Institute Insti tute for Polic Policy y Studi Studies es in Educa Education. tion. Ross, A., & Leathwood, C. (2013). Problematising early school leaving.  European Journal of Education,  48 , 405–418. doi:10.1111/ejed.12038 doi: 10.1111/ejed.12038.. Rumberger, R. W. (1983). Dropping out of high school: The influence of race, sex, and family background.  American  Educational Research Journal,  20 (2), 199–220. doi:10.3102/00028312020002199 doi:10.3102/00028312020002199.. Shavit, Y., & Blossfeld, H. (1993).  Persistent inequality: Changing educational attainment in thirteen countries. ERIC, Social Inequality Series. Boulder: Westview Press. Shavit, Y., & Muller, W. (2000). Vocational secondary education.  European Societies,  2 , 29–50. Solga, Solg a, H. (200 (2002). 2). Stigmatiza Stigmatization tion by negat negative ive selec selection: tion: Explaini Explaining ng less educated people’s decre decreasing asing emplo employmen ymentt opportunities.  European Sociological Review,  18 , 159–178. doi:10.1093/esr/18.2.159 doi:10.1093/esr/18.2.159.. Stearns, E., Moller, S., Blau, J., & Potochnick, S. (2007). Staying back and dropping out: The relationship between grade retention and school dropout.   Sociology of Education,  80 (3), 210–240. doi:10.1177/003804070708000302 doi:10.1177/003804070708000302.. Tawney, R. H. (1951).  Equality. New York: Harcourt, Brace Teese, Tees e, R. (2011) (2011).. Voc Vocati ational onal education education and trai training ning in Fra France nce and Germ Germany any:: Frie Friend nd or foe of the educat education ionally ally disadvan disa dvantage taged? d? In S. Lamb, Lamb, E. Markus Markussen, sen, R. Tees Teese, e, N. Sand Sandberg berg,, & J. Polesel Polesel (Eds.) (Eds.),,   School School dro dropou poutt and  completion: International comparative studies in theory and policy  (pp. 343–357). Dordrecht, Netherlands: Springer. Tumino, A., & Taylor, M. (2013).  The impact of local labour market conditions on school leaving decisions . Paper presented prese nted at Popul Population ation Associat Association ion of Ameri America, ca, New Orleans, Orleans, LA. Van De Werfhorst, H., & Mijs, J. (2010). Achievement inequality and the institutional structure of educational systems: A comparative perspective.  Annual Review of Sociology,  36 , 407–428. doi:10.1146/annurev.soc.012809.102538 doi:10.1146/annurev.soc.012809.102538.. West, A., & Nikolai, R. (2013). Welfare regimes and education regimes: Equality of opportunity and expenditure in the EU (and US).  Journal of Social Policy,  42 , 469–493. doi:10.1017/s0047279412001043 doi:10.1017/s0047279412001043..

 

C o p y r ig n o t b e c e x p re ss in d iv id u

h t o f E u r o p e a n E d u c a t i o n i s t h e p r o p e r t y o f T a y l o r & F r a n c i s L t d a n d i t s c o n t e n t m a y    o p i e d o r e m a i l e d t o m u l t i p l e s i t e s o r p o s t e d t o a l i s t s e r v w i t h o u t t h e c o p y r i g h t h o l d e r ' s   w r i t t e n p e r m i s s i o n . H o w e v e r , u s e r s m a y p r i n t , d o w n l o a d , o r e m a i l a r t i c l e s f o r   a l u s e . 

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