The role of the changing social background composition for changes in inequality of educational opportunity: An analysis of the process of educational expansion in Germany 1950–2010

(2020) - Pia N. Blossfeld

Journal: Advances in Life Course Research
Link:: https://linkinghub.elsevier.com/retrieve/pii/S1040260820300150
DOI:: 10.1016/j.alcr.2020.100338
Links::
Tags:: #paper #Attainment #Education #Germany
Cite Key:: [@blossfeldRoleChangingSocial2020]

Abstract

Using new longitudinal data from the Adult Cohort (SC6) of the National Educational Panel Study (NEPS) this paper examines how the highest educational attainment has changed in the process of educational expansion in Germany between 1950 and 2010. The study focuses on a fairly neglected question of the educational expansion process: Which role does the changing social background composition play in changes in inequalities of educational opportunity? It concludes that, despite a period of substantial educational reform, the overall originspecific inequalities as well as the demand to earn the highest educational degree within the various social origin groups have remained surprisingly stable across birth cohorts. Instead, the general trend towards educational growth at higher education in Germany seems to be mainly driven by intercohort changes in social background composition.

Notes

“This article describes the theoretical reasons and empirical operationalization of a multidimensional social origin measure” (Blossfeld, 2019, p. 1347)

“This social origin variable can be used for the analysis of status inconsistencies as well as cumulating and compensating effects of social origin resources on inequality of opportunity.” (Blossfeld, 2019, p. 1347)

“For example, in educational inequality studies, parental education or parental class or both are often used (e.g. Buchholz and Pratter 2017; Blossfeld et al. 2015; Reimer and Schindler 2010; Müller and Pollak 2010). These uni- or two-dimensional social origin measures are mostly interpreted as interchangeable (Bukodi and Goldthorpe” (Blossfeld, 2019, p. 1347)

“A multidimensional measure provides such a comprehensive understanding of how social origin influences inter- and intragenerational inequality processes through various channels.” (Blossfeld, 2019, p. 1348)

“The interest lies in the decomposition of social origin. An obvious problem of this procedure is that these measures are often quite highly correlated (Blossfeld 2018; Engzell 2016: 2; Betthäuser and Bourne 2016: 15; Chan 2010: 43; see also Table 7 in the “Appendix”). So this approach faces the problem of multicollinearity (Meraviglia and Buis 2015: 38; Marks 2011).” (Blossfeld, 2019, p. 1348)

“However, Chan (2010: 44–49) as well as Blossfeld (2018) could show that there is quite a substantive variability in the distribution of parental education and parental class within parental status groups and vice versa” (Blossfeld, 2019, p. 1348)

“Thus, these status inconsistencies imply that a lack of resources in one family dimension might be compensated by additional resources in another dimension.” (Blossfeld, 2019, p. 1348)

“Other researchers have suggested to pool different family background resources in order to create a latent social origin variable (Barone and Ruggera 2018: 19; Meraviglia and Buis 2015: 37; Buis 2013). Here each of the various social origin resources is theoretically linked to a specific mechanism” (Blossfeld, 2019, p. 1348)

“First, the relationship between the latent social origin measure and the outcome variable is analysed” (Blossfeld, 2019, p. 1348)

“Second, the relative weight of each parental resource is then estimated (Meraviglia and Buis 2015).” (Blossfeld, 2019, p. 1348)

“One advantage of this approach is that multicollinearity is avoided (Meraviglia and Buis 2015: 38). Another benefit is that this latent measure takes into account cumulative effects of parental resources. However, also this approach does not take into account that parental resources can compensate each other (Erola and Kilpi-Jakonen 2017; Saar and Helemäe 2017).” (Blossfeld, 2019, p. 1348)

“The idea of this measure is based on a previous suggestion by Bukodi and Goldthorpe (2013: 1033–1035).” (Blossfeld, 2019, p. 1348)

“It considers cumulative and compensatory effects of different family resources as well as status inconsistencies. Moreover, it is easy to operationalize and it avoids the problem of multicollinearity” (Blossfeld, 2019, p. 1348)

“Sociological theories, whether based on Max Weber or Karl Marx, consider the family as the key unit of social stratification (Erikson and Goldthorpe 1992: 237; Sørensen 1994: 32; Wright 1989). Families are seen as collective entities in which family members share the rewards gained from education, from work and from social deference. As primary agents of socialization, parents are channelling educational, economic and network resources to the next generation. Thus, the family1 is the unit of consumption, where its members live together in a household, where children are raised and where the support of children in education is mainly based (Gambetta 1996: 64; Erikson 1984: 501).” (Blossfeld, 2019, p. 1349)

“there is a study that analysed the individual influence of mother’s and father’s resources on their children’s educational attainment (e.g. Minello and Blossfeld 2014). However, a major limitation of such an individualistic approach would seem to be its failure to acknowledge the role of intimate ties between the partners in conjugal and consensual units as well as the importance of 1 In this article the term “family” is not restricted to conjugal couples with one or more children. Thus, also non-married couples, single parents with at least one child are seen as a family unit.” (Blossfeld, 2019, p. 1349)

“P. N. Blossfeld 13” (Blossfeld, 2019, p. 1350)

“Following Erikson’s (1984, p. 503) idea of the dominance approach, the parental resources of both parents are taken into account to define the multidimensional social origin measure” (Blossfeld, 2019, p. 1350)

“The first family resource that we include in our multidimensional social origin measure is parental education” (Blossfeld, 2019, p. 1350)

“attributional measure, which can be characterized by a distribution in which families with differing amounts of education can be ranked (Goldthorpe and Hope 1972: 23; Goldthorpe 2010: 732). Parental education can be considered as the most important indicator for parents’ ability to support their children’s homework and exam preparation and parents’ familiarity with the educational system which allows them to give more or less informed advice on their children’s educational decisions (Bukodi and Goldthorpe 2013: 3)” (Blossfeld, 2019, p. 1350)

“CASMIN educational scheme” (Blossfeld, 2019, p. 1350)

“The second family resource that we use for our multidimensional social origin variable is parental class (Klasse) (Weber 1976: 177–180). We refer to the narrow concept of class as suggested by Goldthorpe (1996: 487), where parents who dispose of similar economic goods or qualifications and skills that can be translated into income and other returns on the labor market are in the same class (Goldthorpe 1996: 487; Mayer 1977: 159; Müller 1977: 25; Weber 1976). Thus, for our social origin example, we operationalize parental class on basis of the Erikson–Goldthorpe–Portocarero (EGP) class schema (see Sect. 5.2; Erikson and Goldthorpe 1992), which is also implemented in the NEPS” (Blossfeld, 2019, p. 1350)

“Parental status (Stand) is the third family resource that we use to construct our multidimensional social origin variable in our empirical example. This variable is often neglected in many educational studies. It defines the position families have in a hierarchy of social deference (Weber 1976: 177–180). According to Weber (1976) it is another dimension besides class on which society is stratified. Members of each status group (1) seek to associate with superiors and honor them; (2) are in intimate and frequent interaction with equals (such as friends, colleagues and people who often constitute a marriage market); and (3) avoid interaction with inferiors on the prestige hierarchy, which is particularly emphasized by distinct beliefs, values, norms, customs and consumption (Goldthorpe and Hope 1972: 24; Mayer 1977: 155).” (Blossfeld, 2019, p. 1351)

“we use information on the Cambridge Social Interaction and Stratification (CAMSIS) status scale” (Blossfeld, 2019, p. 1351)

“In the context of educational inequality, Bukodi and Goldthorpe (2013: 3) state that parental status represents social and cultural resources that are available to families such as parental social networks and their cultural tastes, which both are seen as important factors for children’s educational opportunities” (Blossfeld, 2019, p. 1351)

“important to stress that in our composite measure of social origin, parental class and parental status are relational concepts (Goldthorpe 2012: 212). This means that they reflect social relations in the labor market and social relations of perceived superiority” (Blossfeld, 2019, p. 1351)

“equality or inferiority that derive from social deference (Goldthorpe 2012: 204, 2010: 732; Goldthorpe and Hope 1972: 23). The positions individuals take in these two kinds of social hierarchies affect children’s educational opportunities and educational choices (Erikson and Goldthorpe 2009; Goldthorpe 2010: 733).” (Blossfeld, 2019, p. 1352)

“it should be noted that there might be further parental resources such as income (Erikson 2016) or wealth (see Müller et  al. 2017 and Weber 1976 for property classes), which might influence family inequality outcomes” (Blossfeld, 2019, p. 1352)

“one can say that the theoretical mechanisms of family income and wealth are quite similar to the mechanisms of social class: economic resources simply make it easier for families to support and help their children in school (Bukodi and Goldthorpe 2013: 2; Erikson 2016). Thus” (Blossfeld, 2019, p. 1352)

“However, on the other hand, income and wealth are what Goldthorpe (2010) calls only ‘attributional’ and not ‘relational’ measures. Thus, social class is a broader concept. It is used to reflect also differences in resources, opportunities and constraints that class members experience in different employment relationships. Thus, parental class (measured in terms of the EGP class scheme) does not only reflect economic inequalities, but in addition covers the influence of social relationships on children’s opportunities.” (Blossfeld, 2019, p. 1352)

“empirical studies demonstrate that parental education is the most important of the three family resources for children’s educational success (Marks 2011; Bukodi and Goldthorpe 2013)” (Blossfeld, 2019, p. 1352)

“It is argued that the three parental resources are highly interwoven and different combinations of parental resources can have compensating and cumulative effects (Erola and Kilpi-Jakonen 2017). Compensation means that families might replace low or intermediate capacities on one of the three resources by higher capacities of the respective other ones” (Blossfeld, 2019, p. 1353)

“Furthermore, accumulation means that a child from a highly educated family is likely to profit also from its high parental class and/or high parental status position.” (Blossfeld, 2019, p. 1353)

“Our suggested multidimensional social origin measure has three further advantages: First, it allows taking into account status inconsistencies. Second, it avoids the problem of multicollinearity that might be present if one introduces the three parental resources separately into the same analysis (see Table 7 in the “Appendix”; Meraviglia and Buis 2015). And third, it is very easy to operationalize” (Blossfeld, 2019, p. 1353)

“dependent dichotomous variable for the grouping of the categories of the social origin measure, an indicator for whether individuals had reached a tertiary education (Y = 1) or not (Y = 0) is used” (Blossfeld, 2019, p. 1354)

“trend variable Our analyses include also a linear cohort trend to model changes in higher educational attainment over time” (Blossfeld, 2019, p. 1354)

“cohort trend variables against birth cohort dummy variables as well as a cohort trend with a quadratic term.” (Blossfeld, 2019, p. 1354)

“To keep the analysis as simple as possible, we use only the linear cohort trend in our analyses (Blossfeld 2018).” (Blossfeld, 2019, p. 1354)

“5.2.1 Parental education As a measure for parental education, we use an updated version of the CASMIN educational schema” (Blossfeld, 2019, p. 1355)

“We use the Erikson–Goldthorpe–Portocarrero class scheme for parental class (Erikson and Goldthorpe 1992).” (Blossfeld, 2019, p. 1355)

“Parental status is measured using the CAMSIS scale (Bukodi et  al. 2011; Chan 2010: 53–54).” (Blossfeld, 2019, p. 1356)

“Since the aim of our illustrative example is to examine combined effects of the three parental resources, the three-category versions of the parental education, parental class, and parental status resources are used to create 3 × 3 × 3 = 27 social origin combinations” (Blossfeld, 2019, p. 1356)

“(1) Social origin combinations 3, 6, 12 and 21 did not empirically exist (see Table 1).” (Blossfeld, 2019, p. 1359)

“For some social origin combinations the number of observations is very small (n < 40), thus no separate statistical analysis is possible for them” (Blossfeld, 2019, p. 1359)

“the categories that were numerically less relevant have been assigned to the respective neighbour categories” (Blossfeld, 2019, p. 1359)

“Then, we ran a binary logistic regression to predict the probability to graduate with a higher education degree (see Table 2). The independent variables are the cohort trend and the 12 remaining social origin groups. We then collapsed the remaining 12 groups into 6 empirically relevant social origin groups (see Tables 2, 3), by combining the social origin groups with similar regression coefficients (see” (Blossfeld, 2019, p. 1359)