Determined to Succeed?

Performance versus Choice in Educational Attainment

Michelle Jackson, Michelle Jackson

• Publisher: Stanford University Press
• Year: 2013
• ISBN:
  9780804784481
• Edition: 1
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

In many countries, concern about socio-economic inequalities in educational attainment has focused on inequalities in test scores and grades. The presumption has been that the best way to reduce inequalities in educational outcomes is to reduce inequalities in performance. But is this presumption correct?

Determined to Succeed? is the first book to offer a comprehensive cross-national examination of the roles of performance and choice in generating inequalities in educational attainment. It combines in-depth studies by country specialists with chapters discussing more general empirical, methodological, and theoretical aspects of educational inequality. The aim is to investigate to what extent inequalities in educational attainment can be attributed to differences in academic performance between socio-economic groups, and to what extent they can be attributed to differences in the choices made by students from these groups. The contributors focus predominantly on inequalities related to parental class and parental education.

CHAPTER ONE

Introduction

How Is Inequality of Educational Opportunity Generated? The Case for Primary and Secondary Effects

Michelle Jackson

When sociologists write about inequalities in educational attainment, they frequently get under way by emphasizing the extraordinary transformations of educational institutions over the course of the 20th century. And indeed, it is hard to imagine how to open a volume on inequalities in educational attainment without acknowledging the significant educational expansion and reform in all Western societies over the last century. The basic features of expansion and reform are well known and may be summarized as comprising three fundamental steps: the establishment of near-universal primary education at the beginning of the 20th century, the rise of near-universal secondary education toward the middle of the century, and the (as yet unfinished) development of a system of mass higher education toward the end of the century. An important consequence of this expansion has been an increase in the average level of educational attainment, such that most of the students who enter secondary education today can expect to obtain a tertiary-level qualification by the end of their educational career. Alongside increasing average levels of attainment, we also observe increasing differentiation in educational systems, so that students may choose from a range of academic and vocational courses in many diverse specialist fields.

The development of educational systems can be understood principally as a response to the demands of changing economic and occupational structures but also as an attempt to create a greater equality of educational opportunity. Yet a great deal of research has demonstrated that significant inequalities in educational attainment between members of different social groups remain. One important area of research focuses on social-class inequalities in educational attainment; children of professional or managerial background generally achieve higher levels of educational performance and make more ambitious educational choices than do children from working-class backgrounds (e.g., Shavit and Blossfeld 1993; Breen et al. 2009).

Arguably of more interest than the current state of class inequalities in educational attainment is the question of how far these inequalities have changed over time. Persistent Inequality (Shavit and Blossfeld 1993) argues that there has been a relatively high degree of temporal stability in the association between class origin and educational attainment. More recent work suggests that a trend toward a weakening association between class origin and educational attainment is present in many European countries, particularly if changes over a relatively long period are considered (Jonsson, Mills, and Müller 1996; Vallet 2004; Breen and Jonsson 2005; Breen et al. 2009, 2010). But this observation should not lead us to lose sight of the following: even if weaker now than in the past, class inequalities in educational attainment remain as a feature of modern societies and this feature is likely to linger for some time. The durability of these inequalities is particularly striking when compared to the far more substantial changes in gender, ethnic, and racial inequalities observed in many countries.

In this volume we aim to understand why social-background inequalities in educational attainment, or inequality in educational opportunity (IEO),¹ should exist and persist in eight Western countries. Should IEO be understood as a consequence of differences in academic ability and performance between members of different social classes? Or should it be understood as a consequence of differences in the educational decisions made by members of different social classes, such that students from advantaged backgrounds choose higher levels of education more frequently than students from disadvantaged backgrounds, regardless of their academic performance? These basic questions outline extreme positions on how IEO is created. In this volume we consider social-background inequalities in educational attainment to be a consequence of both social-background differences in academic performance and social-background differences in the choices that students make, holding performance constant. Our main aim is to determine the relative importance of these two features in creating IEO. Insofar as changes in IEO are observed, we ask whether they can be attributed to changes in the relationship between class and performance or to changes in the class-biased choices that are made, conditional on performance. If we observe declines in one or both effects, leading to a decline in IEO, this provides us with important evidence about which policies and institutional innovations hold most promise for further reducing class effects.

IEO is a term that carries some ambiguity in that it can refer either to a summary measure of all inequalities related to social background generated by an educational system or to only the social-background inequalities generated at a given transition. For example, Boudon defines IEO as meaning “differences in level of educational attainment according to social background” (1974, xi) but also states that “IEO rates are subject to variations as a function of national context, point in time, and school level. . . . [A] certain amount of IEO is present . . . at each school level” (1974, 41). In this volume we are concerned with IEO at given educational transitions, and we therefore take the latter understanding of IEO to be our own. In each chapter IEO is discussed in relation to the transition under consideration and to the risk set of students eligible for that transition. On the whole, we do not address the question of how far IEO assessed for different transitions and risk sets accumulates to a summary measure of all IEO generated by an educational system.²

DEFINING PRIMARY AND SECONDARY EFFECTS

Our understanding of IEO has at its heart an individual-level model, in which a student achieves a certain level of academic performance and then makes a decision about how to proceed in the educational system. The decision that students make when faced with an educational transition is shaped by their previous academic performance, which provides information about the likelihood of successful completion of higher levels of education. But the decision is also influenced by factors other than previous academic performance, because a student takes into account the costs and benefits of the different choices that might be made in relation to the transition.

The decomposition of IEO into a part determined by differences in previous performance across social groups and a part determined by the choices made by members of those groups is well established in the literature, in which performance effects are labeled primary effects and choice effects are labeled secondary effects³ (Girard and Bastide 1963; Boudon 1974). As Breen and Goldthorpe write,

Primary effects are all those that are expressed in the association that exists between children’s class origins and their average levels of demonstrated academic ability. Children of more advantaged backgrounds . . . perform better, on average, than children of less advantaged backgrounds in standard tests, examinations, and so on. . . . [S]econdary effects . . . are effects that are expressed in the actual choices that children, together perhaps with their parents, make in the course of their careers within the educational system—including the choice of exit. (1997, 277)

The history of the conceptual distinction between primary and secondary effects is discussed in more detail below.

Before proceeding any further, an apology on the matter of terminology is in order. While the concepts of primary and secondary effects can be defined with precision, the labels attached to the concepts are unfortunately rather ambiguous. Although one can understand the rationale behind the labels once the concepts have been defined, in that social background primarily affects performance and then secondarily affects choices (conditional on performance), there are few clues to the meanings of these labels for the uninitiated. It is also curious that these terms should be used in a field concerned with education systems, in which “primary” and “secondary” are understood first and foremost to refer to different phases of the school career or school system, not to the decomposition of IEO presented here. Despite the problems with the terminology of primary and secondary effects, the terms are now fully established in the literature on IEO, and in this volume we continue to use these labels. We also refer to primary and secondary effects as “performance” and “choice” effects, so as to further emphasize their meanings. To avoid confusion, we largely avoid using “primary” and “secondary” in isolation to refer to the school career.

THE UTILITY OF PRIMARY AND SECONDARY EFFECTS

There are clear advantages to treating IEO as the overall consequence of the operation of primary and secondary effects: the concepts allow sociologists to gain greater precision in identifying the determinants of IEO, and they also have obvious implications for social policy.

As discussed below, primary and secondary effects are likely to be generated by different processes, and as a consequence, the explanatory tools needed to explain the primary effects of differences in performance between social groups differ from the tools needed to explain the secondary effects of the differences in choices, conditional on performance (e.g., Erikson and Jonsson 1996a). Therefore, by determining to what extent IEO should be understood as resulting from primary or secondary effects, sociologists obtain valuable information about where explanatory effort should be directed. If the distinction between primary and secondary effects is ignored, sociologists are in danger of attempting to explain IEO as an unwieldy whole, thus mistaking a dual phenomenon requiring distinct and separate explanations for a single composite requiring a single explanation.

Aside from the utility of primary and secondary effects for explaining IEO, the concepts are highly relevant to policy. Just as the explanatory tools differ depending on whether the focus is on primary or on secondary effects, the appropriate policy interventions will also differ depending on which type of effect is to be addressed. If differences in performance are the main drivers of educational inequality, policies aimed at reducing those differences will have a very large impact in reducing overall inequality. Early interventions, such as intensive preschool education or economic support for families with young children, seem to have great potential for reducing primary effects (see Cameron and Heckman 1999; Carneiro and Heckman 2003; Heckman 2006). On the other hand, if differences in the choices made by students at the same level of performance have a significant impact, policies aimed at changing constraints and incentives hold more promise (Jackson et al. 2007).

PRIMARY AND SECONDARY EFFECTS

History of the Concepts

The distinction between primary and secondary effects has a long and colorful history. It is a history marked principally by unrealized promise, a state of affairs that can perhaps be attributed to the lack of a rigorous operationalization of the concepts of primary and secondary effects but also to other idiosyncratic features that had a further suppressive effect.

Where to begin with this history? The distinction between primary and secondary effects is most commonly associated with Boudon’s 1974 book Education, Opportunity, and Social Inequality. However, while Boudon’s work is clearly the most extensive treatment of the roles of primary and secondary effects in creating educational inequality, the roots of these concepts are elsewhere. In the 1940s Boalt carried out an empirical analysis of social-class inequalities in educational attainment in Sweden, in which he considered the roles of previous performance and class background in the transition to secondary school (Boalt 1947; summarized in English in Boalt and Janson 1953). Calculating partial correlations between class, school performance, and school selection, on data describing all 5,000 students in primary school in Stockholm in 1936, Boalt concludes that

“social class” measured by father’s occupation gave a significant correlation with marks in primary schools (0.32) but a high correlation with selection to secondary schools (0.56). This latter correlation remained high even when partial correlations for marks, income, and the factor “known to social welfare authorities,” were worked out. (Boalt and Janson 1953, 323)

Boalt clearly distinguishes between the effects of social-class background on performance and its effects on the transition to secondary school, conditional on performance. Here we find, therefore, a distinction between primary and secondary effects in all but name.

It is not apparent that Boalt’s distinction was widely applied in the educational research of the time, and indeed the 1947 book and 1953 paper have been cited only 59 times.⁴ However, Boalt’s work has much in common with that of his contemporaries, in particular, with research in the 1950s and 1960s on the “reserve of talent.” The talent reserve referred to “how many young people . . . with sufficient ability did for various reasons not proceed to upper secondary and university education” (Husén 1974, ix). In this literature, a concern about class inequalities in educational participation rates was expressed as concern about wastage of talent, in that there must be working-class children who were not realizing their full potential. Several studies aimed to assess the size of the talent reserve in relation to class origin, and these studies considered the gap between more advantaged and less advantaged students to be a consequence of differences in ability (or performance) between classes and differences in transition propensities conditional on ability (or performance) (see, e.g., Anderberg 1948; Ekman 1951; Wolfe 1954; Härnqvist 1958; Härnqvist 2003 provides an excellent summary of some of this research). Again, in all but name, these studies distinguished between primary and secondary effects in determining class inequalities in educational attainment. Indeed, the expectancy method, applied by Gösta Ekman (1951), in which the proportion of students from one class making a transition is compared with the proportion of students from another class making a transition for given ability levels, is a precursor of the methods applied in this volume.

In the 1960s the distinction that Boalt had recognized was once again applied in sociological research, although this application was seemingly independent of Boalt’s work. In France, Girard and Bastide published two papers in the journal Population that discussed the effect of social background on performance and its effect on transitions conditional on performance (Girard and Bastide 1963; G...