Social background and school track choice: An analysis informed by the rational choice framework

Abstract

We investigate social inequalities based on social background in the choice of the academic track among equally performing students, and how indicators derived from the rational choice framework contribute to account for such inequalities. We discuss the main theoretical concepts underpinning rational choice theory as applied to educational decisions: perceived costs, benefits, and risks of failure; relative risk aversion; and time-discounting preferences. In the empirical section, we use a unique dataset concerning the transition to different tracks in upper secondary school in a large Southern Italian region. By using various regression methods and the Karlson/Holm/Breen decomposition technique, we show that social inequalities in access to the academic track are considerable, even in recent cohorts, and that they are largely not explained by previous academic performance. Indicators linked to key concepts proposed by the rational choice theory—as measured in this study—account, as a whole, for 31% of the gap based on parental education, and for 40% of the gap based on parental occupation. The most important sources of inequalities among those this study examines are the expected benefits associated with the educational alternatives and the time-discounting preferences, while relative risk aversion and the perceived chances of success play negligible roles.

Keywords

School tracking social inequalities rational choice theory educational expectations social background educational decisions

Introduction

In the field of educational studies, socioeconomic background has been shown to affect children's educational attainment in all industrialized countries. Despite the educational reforms and schooling expansion occurred in all advanced societies, inequalities in educational attainment have only slightly diminished over time and especially in cohorts who experienced the economic boom in the 1960s (Breen et al., 2009). As a consequence, the weakening association between social status and educational attainment over time has not been sufficient to guarantee equality in educational opportunities among recent cohorts (Blossfeld et al., 2016).

In this study, it has become of paramount relevance to improve understanding of the ways in which social background influences students’ educational outcomes. Boudon (1974) provides a useful framework for analysis in this area; it draws a distinction between students’ academic abilities and their educational-related choices. This framework conceives inequalities in educational opportunities (IEO) as the consequence of: first, socioeconomic background differences in students’ academic performance, which comprise the primary effects of social origin; and second, socioeconomic background differences in students’ choices, holding performance constant, which represent the secondary effects of social origin (Boudon, 1974; Jackson, 2013).

The sociological stream of literature that investigates primary effects relies extensively on cultural capital theory (Bourdieu, 1977, 1984; Bourdieu and Passeron, 1990) to explain social inequalities in academic performance (Jæger and Breen, 2016). Secondary effects of social origin have mainly been explained relying instead on rational choice (RC) theory (Breen and Goldthorpe, 1997; Breen and Yaish, 2006; Erikson and Jonsson, 1996; Gambetta, 1987). According to this theoretical framework, the secondary effects of social background are the product of educational decisions families make, that are guided by the expected costs, benefits, and probability of success associated with the school alternatives available at a given educational transition point.

Our study contributes to this second stream of literature by empirically investigating to what extent the arguments derived from the Rational Action Theory—as elaborated by the social stratification literature (Breen et al., 2014; Breen and Goldthorpe, 1997; Breen and Yaish, 2006; Erickson and Jonsson, 1996)—can help in explaining social inequalities in upper-secondary school track choice in Italy, among equally performing students. The secondary effects of social origin are of special relevance in societies that are supposed to rely on a meritocratic system, in which IEO are considered legitimate as long as they are based purely on differences in ability and performance.

We use a dataset derived from a recent survey conducted by Barone et al. (2018a, 2018b) in their longitudinal experimental design. This data source provides rich information about 13-year-old students in the large Southern Italian region of Puglia during academic year 2014–2015, who have been identified as preparing to choose the upper-secondary school track later in the year. By providing a set of indicators capturing key concepts behind RC models, the data source allows us to investigate in detail the reproduction of social inequalities in the most important decision in the educational system (Contini and Scagni, 2013; Gambetta, 1987). The general hypothesis we aim to test is that indicators related to concepts elaborated by the Rational Action Theory mediate the relationship between students’ social origin and secondary school track choice, in particular the propensity to enroll in the academic track.

The Italian context is well suited for this analysis for several reasons. First, the relationship between students’ background and the type of secondary school they choose to attend is comparatively strong (Jackson, 2013) and persistent over time (Panichella and Triventi, 2014). Second, previous performance does not limit students’ choice of type of upper-secondary school; therefore, students and parents are free to make decisions without any formal restriction. This leads to comparatively large secondary effects regarding socioeconomic background (Jackson, 2013). Finally, most of the previous empirical literature has focused on Continental European countries characterized by early and strong tracking (Becker and Hecken, 2009; Holm and Jæger, 2008; Stockè, 2007; Van de Werfhorst and Hofstede, 2007), making it difficult to generalize results to other varieties of educational system. Moreover, a recent Italian study has only focused on access to university, thereby leaving a research gap in providing an explanation for earlier social inequalities in the most important school transition (Barone et al., 2018).

Theoretical framework

RC theory and the Breen–Goldthorpe model

The idea that educational choices can be the result of individual rational decisions and evaluations was developed for the first time in contraposition to the structuralist view that conceives individual actions as determined by external constraints (Elster, 1979). RC theory views individuals as acting “purposively in accordance with their intentions” (Gambetta, 1987: 16). This means that they can rationally evaluate their present situation in order to maximize returns later on, evaluating costs and benefits of specific choices. This framework is particularly suited to understanding educational choices, conceived as forms of investments: the estimated utility of a given educational path for a student can be understood as a function of expected costs, probability of success and expected benefits (Erikson and Jonsson, 1996). The costs of an educational pathway can be both direct—involving fees for tuition, books, and materials—and indirect—such as foregone earnings. The second parameter for students’ estimation of utility is their probability of success, or their expected likelihood of successfully completing a given educational path, which is strongly influenced by students’ previous academic competence (Breen and Goldthorpe, 1997). Lastly, parents and students have specific ideas about the benefits associated with the successful completion of a determined educational career, mainly in terms of monetary returns or occupational prestige.

Breen and Goldthorpe (1997) integrate this general framework with the concept of relative risk aversion, conceived as the most important criterion in guiding educational choices among students with the same academic performance.¹ According to the Breen–Goldthorpe model, the leading motive driving families’ rationally informed educational choices is the desire to avoid downward mobility for their children. However, the aim of maintaining their occupational status means different things for families located in the different social strata. Although for working-class students, achieving an upper-secondary diploma translates to significantly increased chances of avoiding downward mobility, this is not true for upper-class children, who need to obtain a tertiary degree to maximize their chances to preserve their parents’ social standing (Obermeier and Schneider, 2015). The Breen–Goldthorpe model predicts, therefore, that children from socioeconomically advantaged families are more likely to make ambitious educational choices, for example, attaining a university degree, than their counterparts from less advantaged families, even when they have the same academic achievement.

More recent developments suggest an additional mechanism for explaining educational choices within an RC framework. Relying on the concept of time preferences developed by economists Becker and Mulligan (1997), Breen et al. (2014) have elaborated a model for educational decisions in which the utility of such decisions is determined by risk aversion and time discounting preferences. Time discounting preferences refer to the extent to which students attribute different values to immediate returns than to returns occurring later. Students with short time horizons will be more likely to opt for vocationally oriented paths, while those with longer time horizons will be more prone to select an academic track that leads to higher education. Hillmert and Jacob (2003) operationalize the expected returns in terms of future income within a certain time—time horizon— and treat it as a key parameter in the theoretical model for analyzing social inequalities in higher education.

Even if educational choices are conceived as rational calculations among costs, benefits and possible alternatives, this does not exclude an a priori intervention of cultural influences in a context of limited rationality (Simon, 1955). Choices are also conditioned by the individual's socioeconomic class, and corresponding specific opportunities, constraints, and beliefs. Consequently, rational evaluations are understood to be driven by individual perceptions about risks, costs and benefits, together with socially stratified information biases (Barone et al., 2018a, 2018b).

The Italian context

The Italian education system

In the past 50 years, the Italian schooling system has undergone several de-tracking reforms that reduced barriers in access to education by limiting its stratification. These reforms aimed to increase participation in all levels of education, reducing social-background inequalities (Cobalti and Schizzerotto, 1994). Students enter the school system at age 6 years and attend 8 years of comprehensive education. At the end of lower secondary school, students can choose among several upper secondary school types with different educational programs. The variety of upper secondary schools can be classified broadly into the academic track (licei) and two non-academic tracks: the technical track (istituti tecnici) and vocational track (istituti professionali). The three tracks have heterogeneous institutional purposes, academic standards, curricula, and prestige (Contini and Triventi, 2016; Gambetta, 1987).

In general, lyceums are considered to be the main path on the way to university. In contrast, technical schools combine general education with a vocational education, and are usually considered to be less demanding. Technical institutes provide competencies and skills in economic or technological sectors, and prepare students for technical jobs. Vocational education lasts from 3 to 5 years and provides practical knowledge for lower-level practical and technical occupations. After 5 years of schooling, all the students take a national examination (esame di maturità), but the final evaluation is not determinant in eventually continuing to a tertiary degree. Indeed, there are no entry restrictions to university, although some degree programs admit a closed number of students per year and thus require ability-based entry tests that can be more or less demanding.

Patterns of educational inequalities in Italy

Despite the long-term, generalized, all-encompassing, and notably strong decrease in inequalities among educational opportunities in Italy (Barone et al., 2010; Triventi et al., 2016), inequalities in educational opportunities persist and continue to affect the educational paths of young Italian students. Concerning vertical inequalities, research has thoroughly demonstrated that students with higher social background are more likely to reach the highest educational levels (Ballarino et al., 2009; Triventi and Trivellato, 2009). Concerning horizontal inequalities, students with a higher social background are also more likely to attend more demanding and higher-quality educational programs (Checchi and Flabbi, 2013). The relationship between the type of secondary school attended by Italian students and their socioeconomic background is robust and persistent over time (Panichella and Triventi, 2014), with only slight decreases in the most recent cohorts (Guetto and Vergolini, 2017). The passage from lower secondary education to upper secondary school is a key transition in the Italian educational system, with longer-term consequences for student access and success in tertiary education. Indeed, inequalities in university enrollment and persistence based on socioeconomic background are driven largely by differences in upper secondary schooling paths (Argentin and Triventi, 2011; Ballarino and Panichella, 2016; Contini and Triventi, 2016).

Due to the lack of suitable data, only a few studies have ventured into the decomposition of primary and secondary effects. Results show that overall, in Italy, secondary effects account for between 50% and 60% of students’ differentials in educational transition probabilities due to socioeconomic background (Contini and Scagni, 2013; Contini and Triventi, 2016; Ress and Azzolini, 2014). As Jackson (2013) demonstrates, this share is rather high in comparative perspective.

With the exception of a few earlier studies (Gambetta, 1987; Schizzerotto, 1997), only one recent study has aimed to test the role of RC mechanisms in explaining secondary effects in university enrollment, using a rich set of variables (Barone et al., 2018a, 2018b). They found that the most important factors are the indirect assessment of costs, perceived chances of success, and occupational returns; and that relative risk aversion has only a modest explanatory power. Overall, the mechanisms of RC theory mediate between 18% and 25% of the secondary effects of socioeconomic background in university enrollment. However, these results reflect social inequalities in a rather late educational transition, in which the students have already been exposed for five years to a tracking system. This leaves open the question of whether decisional mechanisms postulated by RC theory have a stronger role at lower educational stages, especially in the first transition to heterogeneous educational programs.

Research design

Data

We use a dataset that comes from the longitudinal experimental survey conducted by Barone et al. (2018a, 2018b). The original sample includes 3113 students preparing to leave lower secondary school, in 157 classes, selected in Puglia, a large Southern Italian region. The sample of the survey was selected by a random, two-stage sampling design, proportionally stratified by province, using 44 lower secondary schools. Depending on the size of the school, up to four classrooms for each school are included. In October (2014), at the beginning of the final year of lower secondary school, all the sampled students answered a questionnaire about their intended future educational choices; scholastic performance; school and classroom environment; relationships with peers and teachers; family environment; and socioeconomic background. The response rate at this stage of the research was 87%. One year after the initial survey, a follow-up CATI interview with the students asked for students’ effective track choice, but at the first stage the sample was split in two because an information experiment on a selected target took place (Barone et al., 2018a, 2018b). For this reason, we have used educational expectations as our main outcome, and have conducted an additional robustness check analyzing actual track choice on the subsample of non-treated students in the Online Appendix (Section B).

We rely on this data source since it provides a number of variables ascribed to the RC framework and covering the transition to upper secondary school. The use of “local” samples tends to be the norm in this stream of the literature, and despite this dataset not being formally representative at the national level, additional analyses conducted on national administrative data suggest that the mechanisms for reproduction of educational inequalities work very similarly across Italian regions. After casewise deletion, the analytical sample includes 2448 observations (see Online Appendix Table 3 for information on missing cases).

Variables

Our dependent variable is the intention to attend an academic track, as provided by students in their final year of lower secondary education (eighth grade).² The variable is derived from the student's declaration about the specific curriculum they aim to attend in high school, which was recoded into a dummy variable indicating lyceum versus technical and vocational schools.³ We focus on this dichotomous variable for various reasons: (1) it reflects the most consequential choice in terms of subsequent outcomes in education and in the labor market (Barone et al., 2021; Triventi et al., 2021); (2) it increases the comparability with the established research that uses a similar definition (e.g. Stockè, 2007; Van de Werfhorst and Hofstede, 2007); (3) the indicators we use are mostly referred to this distinction; (4) empirically, the strongest inequalities are found in the choice between the academic track and the vocationally oriented tracks. Additional analysis with a tripartite distinction of tracks is reported in Section E of the Online Appendix.

Social background is captured by two variables indicating the highest parental education and parental occupation, with each variable having four categories. Pre-tracking academic performance is measured by computing the average mark among the most important subjects; it ranges from 1 to 10, with six representing the passing mark. Additional control variables are gender, migration background, and regularity in school career (see Table 1).

Table 1.

Variables description (analytical sample N = 2448).

Variable	Variable description
Outcomes
Intended track choice	(0) Technical school or vocational school; (1) Traditional lyceum or other lyceums
Independent variables
Parental occupation	(1) Lower (farmers, workmen or un-qualified employees); (2) Medium-low (owners and retailers); (3) Medium-high (expert or qualified employees); (4) Higher (executive or manager, professional and entrepreneur). Highest job position between the mother and the father is selected
Parental education	(1) Lower secondary; (2) One with upper secondary; (3) Both with upper secondary; (4) University. Composite variable created using both indicators of mother education level and father education level
Academic performance	Range 1–10, where 6 is the passing mark. Mean between different subjects (native language, foreign language, mathematics, science)
Rational choice indicators
Perceived chances of success (P1)	“How much do you think you are cut out for studying?” Scale: 1 (not at all) to 10 (a lot)
Perceived risk of failure (P2)	“In choosing your future type of school, how much important is the possibility of attending a school that is not too difficult and where you expect to succeed” Scale: 1. Not important at all; 2. Little important; 3. Quite important; 4. Very important (Standardized variable)
Perceived benefits 1 (B1)	“How important is the chance to find a good job with the diploma without going to the university in choosing your future type of school?” Scale: 1. Not important at all; 2. Little important; 3. Quite important; 4. Very important (Standardized variable)
Perceived benefits 2 (B2)	“How important is receiving a good preparation in order to go to University in choosing your future type of school?” Scale: 1. Not important at all; 2. Little important; 3. Quite important; 4. Very important (Standardized variable)
Perceived risk of social demotion (RRA)	“How important is the possibility to do the same job as your parents in choosing your future type of school?” Scale: 1. Not important at all; 2. Little important; 3. Quite important; 4. Very important (Standardized variable)
Time discounting preferences (TDP)	“How important is the opportunity to go to work as soon as possible in choosing your future type of school?” Scale: 1. Not important at all; 2. Little important; 3. Quite important; 4. Very important (Standardized variable)
Economic wealth (C)	“Regarding the following items, can you tell if you have it at home and, if you have, how many?” A. Mobile phone; B. Television; C. Personal computer; D. Car; E. Washing-machine. Scale: 1. None; 2. One; 3. Two; 4. Three or more. Single items were recorded in dummy variables (0 = relatively lower econ. wealth; 1 = relatively higher econ. wealth). Additive standardized scale.
Socio-demographic control variables
Gender	(0) Male; (1) Female
Country of birth	(0) Italy; (1) Abroad
Regularity in studies	(0) Regular students; (1) Irregular students

Measuring the concepts outlined by RC theory is a challenging task, as no consensus has yet been reached on the best empirical solution. Moreover, relying on secondary data imposes further limitations, given the impossibility of designing new questions specifically tailored to this end. However, our data are derived from a survey designed by social stratification researchers, and contain a set of questions capturing key concepts of RC theory.

We consider two variables measuring the perceived benefits, and two measuring the expected probability of success or risk of failure. Relative risk aversion, intended as desire to avoid social demotion, is captured by a variable indicating how important is for the respondent, when choosing the high school track, to have the opportunity to perform the same job or career as the parents. While this variable is imperfect to capture the breadth of relative risk aversion, it is the best available alternative in the data, and shares significant commonalities with indicators used in previous works (Becker and Hecken, 2009; Stockè, 2007). Taking into account other studies (Breen et al., 2014), we also include an indicator that can be considered a proxy for students’ time discounting preferences; it indicates the extent to which respondent places a high value upon the opportunity to go to work as soon as possible when choosing the high school track.

It is important to bear in mind that, among the selected indicators, some are more related to the expectancy-components of RC, for example, success probability, whereas others are more related to the value component (importance), for example, the expected benefits. The selected indicators show moderate to low degrees of correlation with each other (see Table A1, Online Appendix).

Analytical strategy

Figure 1 depicts our conceptual scheme guiding the empirical analysis. The empirical analysis is divided into four steps, which are the analysis of: (1) the association between social background and intended track choice; (2) the relationship between social background and RC indicators; (3) the relationship between RC indicators and track choice; and (4) the role of RC indicators in mediating the relationship between social background and track choice, conditional on academic performance.

1) The first step concerns the analysis of the social stratification of track choices in secondary school. By using binomial logistic regression models, we analyze how students’ social background is related to their intention to enroll in the academic track, AT. The models are specified as follows:

M_{1} : A T = f (S o c B a c k | S o c i o D e m o)

M_{2} : A T = f (S o c B a c k | S o c i o D e m o, A c a d P e r f)

Figure 1.

Conceptual model beyond the analytical approach.

Each model is repeated twice, because we include indicators for social background separately in order to avoid collinearity and measure the gross effect of each independent variable. Additional models relying on a comprehensive indicator of social background are reported in Section F of the Online Appendix. Moreover, we employ the Karlson/Holm/Breen (KHB) technique (Karlson and Holm, 2011) to decompose the total effect of social origin into a direct and indirect effect via academic performance.⁴ This allows us to assess the extent to which previous academic performance accounts for social origin effects on the expected track choice (primary effects), and to assess the residual effect of social background (secondary effects).

2) The second step examines the relationship between social background and the selected RC indicators through different linear regression analyses, which adjust for students’ sociodemographic characteristics.

M_{3 - 9} : R C I = f (S o c B a c k | S o c i o D e m o)

3) The third step investigates the association between the indicators of RC theory and intended track choice. This is done through a binomial logistic regression predicting the probability of choosing the lyceum, controlling for all variables in Table 1.

M_{10} : I T C = f (R C I | S o c B a c k, S o c i o D e m o, A c a d P e r f)

The second and third steps are important in view of the last analysis, since RC indicators can explain social inequalities in track choices only if they are socially stratified and, at the same time, they affect the choice of the academic track.

4) The fourth step assesses the role of the RC indicators in explaining social inequalities in students’ intended upper secondary school choice. We use the KHB technique to decompose the separate effects of parental occupation and parental education on the expectation of enrolling in the academic track, net of students’ sociodemographic characteristics and students’ academic performance. This model is calibrated on the theoretical idea outlined in Breen and Goldthorpe (1997), in which RC aims to explain why upper-class children are more likely to choose the more ambitious educational pathway compared to their equally skilled lower-class counterparts. Most results are reported in graphical form in the article, in order to ease the interpretation of substantive results.⁵ Given that, as most studies in this area, we rely on cross-sectional data, to avoid issues of reverse causality we need to assume that the individual expectations and value attributed to costs, benefits, and probability of success are developed in advance and actually affect the decision to enrol in the academic track—and not the opposite.

Results and findings

Intention to enroll in the academic track by social background

The first step is analyzing the extent to which parental occupation and parental education are associated with students’ intention to choose lyceum instead of a technical or vocational track. The dark gray bars in Figure 2 show the proportion of lower secondary students who have the intention to choose lyceum, predicted by binomial logistic regression models. Graph A shows that the predicted probability of choosing lyceum is 0.40 among students with lower parental occupations, while it amounts to 0.69 among students coming from higher occupational classes. A similar pattern is encountered when looking at parental education (Graph B): choosing lyceum is less common both for students whose parents have a lower secondary degree (0.35), and for students with only one parent with an upper secondary degree (0.48); choosing lyceum is significantly more common among students whose parents both have an upper secondary degree (0.63), and among students who have at least one parent with a university degree (0.77).

Figure 2.

Binomial logistic regressions predicting the intention to enroll in lyceum by parental occupation (graph A) and by parental education (graph B); average predicted probabilities; 95% confidence intervals of parental occupation and parental education; model 1 only for students’ social background; model 2 controlled for students’ academic performance and sociodemographic characteristics (gender, country of birth, regularity in studies); N = 2448. Note: Clustered standard error at the school level. Model 1 reports the total effect, model 2 the effect adjusted for previous academic performance.

The light gray bars in Figure 2 show the same average predicted probabilities adjusted for academic performance; these should be interpreted as the expected likelihood of opting for lyceum by parental education and occupation for students with the same academic performance before tracking. Even after controlling for students’ sociodemographic characteristics and their previous academic performance, the relationship between social background and intended upper secondary school choice is maintained and remains strong.

In order to establish, in a more formal way, to what extent the higher propensity to enroll in the academic track displayed by high-status students is due to their superior academic achievement, we performed a KHB decomposition. Table 2 reports the estimates of the total, direct, and indirect effects, expressed as average partial effects. The last two columns present the estimates of primary and secondary effects in percentages relative to the total social origin effect. Regarding the intention to enroll in lyceum as opposed to technical or vocational schools, secondary effects account for 50% of students’ inequalities due to parental occupation, and for 60% of students’ inequalities based on parental education. These results, albeit obtained from a specific Italian region, are fully in line with previous results on the entire country reported in previous studies at the national level (Contini and Scagni, 2013; Contini and Triventi, 2016).

Table 2.

KHB decomposition predicting the intention to enroll in lyceum: average partial effects for direct, indirect and total effect and primary and secondary effects of parental occupation and parental education, and corresponding statistical significance levels (*p < 0.05, **p < 0.01, ***p < 0.001); N = 2448.

Intention to enroll in lyceum	Direct(full model)	Indirect(est diff)	Total (reduced model)	Primaryeffects	Secondary effects
Parental occupation
Medium-low	0.0337	0.0503	0.0840**	60%	40%
Medium-high	0.136***	0.115	0.251***	46%	54%
Higher	0.162***	0.129	0.292***	44%	56%
Simple average	–	–	–	50%	50%
Parental education
One with upp sec	0.0794***	0.0489	0.128***	38%	62%
Both with upp sec	0.159***	0.121	0.280***	43%	57%
University	0.251***	0.167	0.418***	40%	60%
Simple average	–	–	–	40%	60%

Note: clustered standard errors at the school level; KHB decomposition with concomitants (gender, country of birth, regularity in studies). Primary effects are computed as ((Indirect/Total)/Total)*100; secondary effects are computed in a residual way as 100-Primary effects.

The social stratification of indicators related to RC theory

The further step is to analyze whether the indicators selected to measure key RC theory concepts are related to students’ social background. Figure 3 reports coefficients derived from several linear regression analyses, predicting the expected value of each indicator by parental occupation and parental education separately, adjusting for students’ sociodemographic characteristics. To simplify the interpretation, we focus on the comparison between the two extreme categories of the social background variables, and report the coefficients following their effect size (Online Appendix Figure 1 reports the full set of coefficients).

Figure 3.

Linear regression analysis predicting rational choice indicators; 95% confidence intervals of parental occupation (graph A) and parental education (graph B); N = 2448. Note: Clustered standard errors at the school level; all standardized indicators; coefficients controlled for students’ sociodemographic characteristics (gender, country of birth, regularity in studies).

RC indicators, overall, are clearly socially stratified, and the patterns are similar upon examination of the two dimensions of social origin. Students from different social backgrounds have significantly different perceptions of the benefits, costs, and probability of success related to the educational options available after lower secondary education in Italy. The differences are not only statistically significant for all the RC indicators, but for most of the indicators are also substantially relevant, as they sit between .25 and .70 standard deviations, with slightly larger estimates on the parental education indicator.

Social background is more strongly related to students’ perceived benefits. When choosing the school track, students from higher social backgrounds are much less likely to consider the chance of finding a good job without attending university to be important. Moreover, they are also much more likely to opt for a school that guarantees better academic preparation and increases the chances to successfully attend university.

As expected, social background is also strongly related to the economic wealth of the family, with upper-class children reporting higher numbers of items related to material wealth at home. Social origin, net of other variables, is also related to the perceived chances of success. Upper-class children and those from tertiary-educated families are more prone to consider themselves cut out for studying, and less likely to avoid demanding curricula.

Despite being overlooked by most previous studies, students’ time discounting preferences also appear to be related to social origin. Students with high social backgrounds are less likely to consider the opportunity to go to work as soon as possible to be an important criterion in choosing the future type of school, compared to their lower-background counterparts.

Finally, in line with previous studies (Stockè, 2007; Van de Werfhorst and Hofstede, 2007), our proxy for relative risk aversion is only moderately stratified. Students from more educated families are more likely to aim for the same profession as their parents, compared to lower-class students. However, the differences between social groups are modest in size compared to those observed for the other RC indicators.

RC theory indicators and intention to enroll in lyceum

The third step of the analysis is assessing the relationship between RC indicators and the intention to enroll in the academic track. In Figure 4, we report the average marginal effects from binomial logistic regression, predicting the probability of having the intention to choose the academic track after lower secondary education.

Figure 4.

Binomial logistic regression predicting the intention to enroll in lyceum by RCT indicators: average marginal effects; 95% confidence intervals; N = 2448. Note: Clustered standard error at the school level. Average marginal effects controlled for students’ social background (parental education and parental occupation), students’ sociodemographic characteristics (gender, country of birth, regularity in studies) and academic performance.

In contrast to the previous analysis, in which RC indicators were all socially stratified in the expected directions, we notice that only some of the indicators are related to track choice intentions. The perceived benefits are again the most important variables: an increase in these indicators strongly increases the chance to opt for the academic track.

Time discounting preferences, perceived benefits and perceived chances of success have a statistically significant effect on planning to attend lyceum. Students who aim to find a good job with the upper secondary diploma without going to the university are less likely to opt for the lyceum (−6 p.p.), while those who seek a school that enhances academic preparation and the chances to succeed at university are more likely to opt for the lyceum (9 p.p.). Moreover, students who think they are cut out for studying are slightly more likely to opt the academic track, but the effect is much smaller than the one related to perceived benefits (2.5 p.p. for a standard deviation increase in the index). In contrast, students who attribute a high value to the opportunity to commence working to earn money as soon as possible, are less likely to intend to enroll in the lyceum (−5 p.p.). Results are obtained by comparing equivalent students in terms of social background, sociodemographic characteristics, and academic performance, and including all RC indicators together. Once all the variables are included in the models, relative risk aversion, economic wealth (proxy of expected costs), and perceived risk of failure are shown to be unrelated to the track choice intention.

Social inequalities in the intention to enroll in lyceum: The contribution of RC indicators

In Table 3, we decompose the secondary effects of social background—the effect of social origin among equally competent students (reduced model)—into a part that is explained by RC indicators (indirect effect) and a residual unexplained part (direct effect). The indirect effect is presented both in absolute and in relative terms (percentage of the total effect). For the sake of brevity, we focus on the comparison between extreme categories of social background, as performed in the previous analyses.

Table 3.

KHB decomposition predicting the intention to enroll in lyceum: average partial effects for direct, indirect and total effect of parental occupation and parental education controlling for RCT indicators (and academic performance as a concomitant), and corresponding statistical significance levels (*p < 0.05, **p < 0.01, ***p < 0.001); N = 2448.

Intention to enroll in lyceum	Direct(full model)	Indirect(est diff)	Total (reduced model)	Indirecteffects %
Parental occupation
Medium-low	−0.001	0.027	0.026	–
Medium-high	0.085***	0.046	0.131***	35%
Higher	0.097***	0.064	0.162***	40%
Simple average	–	–	–	37%
Parental education
One with upp sec	0.052*	0.026	0.078***	34%
Both with upp sec	0.105***	0.051	0.156***	33%
University	0.172***	0.079	0.250***	31%
Simple average	–	–	–	33%

Note: clustered standard errors at the school level; KHB decomposition with concomitants (academic performance, gender, country of birth, regularity in studies). Indirect effect in percentage on the comparison between medium-low and low occupation is not computed since the total effect is not statistically significant and negligible in size.

Among equally competent students, those with tertiary educated parents are, on average, 25 p.p. more likely to attend the lyceum compared to those with parents with no more than lower secondary certificate. The mechanisms identified by RC theory—as measured in this study—accounts as a whole for 31.4% of the gap. The difference in the predicted probability of intending to enroll in the lyceum is instead lower when looking at the comparison between higher and lower parental occupation (16 p.p.). However, RC indicators account for a larger part of social inequalities, since the share of the explained gap is almost 40%.

Figure 5 reports the results of the detailed decomposition, in which we assess the contribution of each single RC indicator in explaining the effects of social inequalities on the intentions related to track choice. The results are generally straightforward, and two factors have significant explanatory power: (1) the perceived benefits associated with educational alternatives in high school (good preparation for university studies, qualification that allows to enter immediately in the labor market); and (2) time discounting preferences.

Figure 5.

KHB decomposition predicting the intention to enroll in lyceum by social background (higher vs. lower); coefficients and 95% confidence intervals for rational choice indicators; N = 2448. Note: Clustered standard errors at the school level; KHB decomposition with concomitants (academic performance, gender, country of birth, regularity in studies).

In absolute terms, the indirect effects associated with these indicators are highly similar when parental education or occupation is considered. In relative terms, perceived benefits account for 23% of the gap between tertiary- and lower-secondary-educated parents, and 29% of inequalities between higher and lower occupational groups. Time discounting preferences account for smaller portions of secondary effects, between 4% and 6%. The contribution of the other RC factors (economic wealth as a proxy of costs, perceived risk of failure and chances of success, relative risk aversion) is instead not statistically significant, and substantially small. As reported in Section C of the Online Appendix, the importance of RC indicators does not substantially change if we include additional indicators that possibly explain social inequalities, such as variables of sociocultural resources or students’ attitudes. Although focusing on the respective contributions of single RC indicators can provide some useful insights, some of the differences among their importance might stem from the measurement quality of each variable; they should thus be interpreted with caution.

Discussion and conclusion

In this article, we have aimed to investigate whether a set of indicators derived from the RC theory, especially those elaborated in the social stratification literature (Breen et al., 2014; Breen and Goldthorpe, 1997; Breen and Yaish, 2006), can account for social inequalities in enrollment in the academic track, a crucial educational decision in the Italian educational system. The sociological version of RC theory proposes that educational decisions are driven by considerations of the costs, benefits, and expected probability of success associated with each educational alternative at a specific school transition point. While earlier studies have focused on the decision to enroll in a given educational level once the previous one has been completed (i.e. Breen and Goldthorpe, 1997), the decision process is equally applicable to horizontal educational decisions, in which students are asked to decide about the specific track, curriculum or field of study (Gabay-Egozi et al., 2010; Stockè, 2007). This is particularly relevant in those countries in which almost all students participate in lower educational levels, and in which the qualitatively different types of education became increasingly important in affecting later educational transitions (Blossfeld et al., 2016; Lucas, 2001; Triventi et al., 2019).

To this purpose, we contributed to the literature by analyzing the transition of a recent cohort of students in a large Southern Italian region to upper secondary education, a key passage in the Italian education system. It is indeed the first selection point into different educational branches, and the track attended is strongly related to successive dropout risk; retention likelihood; the probabilities of entering university and completing a degree; and longer-term occupational returns (Ballarino and Panichella, 2016; Barone et al., 2021; Contini and Triventi, 2016).

From an international viewpoint, RC theory has been tested mostly in Continental European countries, such as Germany, the Netherlands, and Switzerland; these studies have characterized mostly by early selection into tracks and a comparatively strong role of teachers in the sorting process (Becker and Hecken, 2009; Stockè, 2007; Van de Werfhorst and Hofstede, 2007; Zimmermann, 2020). Italy is a strategic case study, because selection into tracks occurs later (around age 14 years) and teachers do not play a relevant role in students’ destinations; therefore, in Italy, students have more voice in the decision. Compared to previous research, we have also attempted to provide a more comprehensive investigation of the role of indicators related to concepts elaborated by RC theory, by considering multiple indicators, where available, analyzing expected costs, benefits, and probability of success, and also considering role-status maintenance motives and time-discounting preferences.

In line with previous results, we have found that higher-background students are more likely to opt for the academic track than for technical and vocational schools, compared to lower-background students. The unconditional differences are significant, ranging from 29 (parental occupation) to 42 percentage points (parental education). The differences are in part due to heterogeneous levels of pre-tracking academic performance, but this factor accounts for less than half of the gap. This means that a large segment of inequalities in access to the academic track occurs among children with similar academic performance but different socioeconomic resources at home, which Boudon (1974) has called the “secondary effects” of social origin.

Among students with the same academic performance, the conditional differences in intention to attend the academic track remain large for parental education (25 percentage points comparing tertiary and lower educated parents), and are smaller but nevertheless significant for parental occupation (10 percentage points). The indicators related to concepts elaborated in the RC theory—as measured in this study—account together for 31% of the gap based on parental education, and for 40% of the gap based on parental occupation.

In comparison to the work of Barone et al. (2018), which found RC indicators explaining less than a quarter of the differences in university enrollment linked to socioeconomic background (among equally competent students), our indicators display larger explanatory power, despite some of our variables possibly not capturing RC concepts as well as those used in the previous study. For this reason, the share of inequalities statistically accounted for by the RC indicators could be tentatively considered a lower bound estimate in our work. As a consequence, our indicators display stronger explanatory power in earlier school transitions in which students are not already separated into differentiated tracks, characterized by different student body composition. However, studies that are able to compare the relative explanatory power of RC theory across educational transitions using more comparable data and indicators are currently lacking, which suggests a possibility for further development in this field of research.

We have also found heterogeneous contribution for each RC indicator to explain social inequalities in the intention to attend the academic track. In the Italian context, secondary effects of social origin in the transition to upper secondary school are to a large extent driven by students’ expected benefits and time-discounting preferences. Thus, a key driver of inequalities in the first allocation into differentiated tracks is the extent to which students place importance upon improving their chances to find a good job with an upper secondary diploma, or, on the contrary, find importance in receiving a preparation that will maximize their chances of successfully attending university. However, the decisions related to these beliefs appear to be misplaced in the Italian context: occupational prospects are in any case improved for those who attain an academic qualification, even among those who do not graduate from university (Barone et al., 2021).

The other RC indicators, instead, do not contribute sufficiently significantly to account for the gaps between higher- and lower-background students. Our indicator of status maintenance motives is weakly related to social origin, and does not independently affect the probability of choosing the academic track. While our measure is clearly imperfect, this result echoes those found by previous studies that employed different and more elaborated measures (Barone et al., 2018; Stockè, 2007; Van de Werfhorst and Hofstede, 2007), in which relative risk aversion displays a modest explanatory contribution of inequalities as well.

Finally, it behooves us to discuss some limitations and avenues for future research. The primary limitation is that, given the available data, we were able to provide only an empirical assessment of the statistical importance of various indicators we presume are connected to key RC concepts, as elaborated in the social stratification literature, in accounting for social inequalities related to upper secondary track choice. We were not able to directly test the mechanisms proposed by a more comprehensive version of the RC theory, which would require a formal comparison of the Subjective Expected Utility across the full set of educational alternatives (see Becker and Glauser, 2018). Additionally, to avoid issues of reverse causality, it would be important to rely on longitudinal data with repeated measures of the RC indicators. Future studies should address these important issues, to further improve the empirical assessment of the role of RC theory in explaining social inequalities in educational decisions.

A second limitation involves the operationalization of our indicators. Although the questionnaire was designed by social stratification researchers, the main aim of the survey was different from our goal (see Barone et al., 2018a, 2018b), so we were constrained in the choice of indicators to capture RC parameters. We attempted to measure the same concept relying on more than one variable (i.e. expected benefits and probability of success), but this was not feasible for all the indicators, and were not in the position to use indicators for both the value and expectancy components of each RC parameter.

It is further possible that in this study some indicators are better measured than others. For example, we do not provide any subjective indicator of indirect and direct costs, but only a family wealth indicator. Even if socially stratified in the expected direction, this indicator seems not directly related to intention of enrollment in the academic track. Based on results by Barone et al. (2018), we could speculate that a better measure of the perceived (indirect) costs, would have contributed to explaining IEO to a more detailed extent. Moreover, our indicator of relative risk aversion is poorly measured, and likely cannot adequately convey the complex notion of status maintenance. Future researchers should devote more attention to the measurement of RC concepts and find consensus on the best practices, after having validated composite indicators in line with what is done in the psychometric literature.

Finally, it is noteworthy that we have implicitly considered the students as the pivotal actor in track choices, registering their beliefs and perceptions about the alternative choices. One consideration is that in Italy the track choice happens when students are older than in Continental European countries, so it is likely they have developed their own opinions. Conversely, another consideration is that parents might have a strong role in educational-related decisions, especially at lower educational levels (Müller and Karle, 1993). In cases where parents’ and children's perceptions are not aligned in perceptions of costs, benefits, and risk of failure, and where actual choices are driven more by parents than by students, our findings would tell only a part of the story. These issues should be taken into account in future surveys on education-related decisions, in which it would be optimal to measure RC-related perceptions for both parents and children, in order to more clearly disentangle the drivers of the reproduction of educational inequalities.

Supplemental Material

sj-docx-1-asj-10.1177_00016993211061669 - Supplemental material for Social background and school track choice: An analysis informed by the rational choice framework

Supplemental material, sj-docx-1-asj-10.1177_00016993211061669 for Social background and school track choice: An analysis informed by the rational choice framework by Ilaria Lievore and Moris Triventi in Acta Sociologica

Footnotes

Acknowledgements

The authors thank Carlo Barone for sharing the data and for providing valuable comments on our work. An earlier version of this paper was presented at ISA-RC28 Spring Meeting in Frankfurt (2019). They thank the participants who gave us useful insights on that occasion. Usual disclaimers apply.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Moris Triventi acknowledges funding support by the Compagnia di San Paolo Foundation for the INEQUALITREES project (2021.AAI274.U302), as part of the 2018 “Global Issues – Integrating Different Perspectives on Social Inequality” call.

ORCID iD

Ilaria Lievore

Supplemental material

Supplemental material for this article is available online.

Notes

Author biographies

Ilaria Lievore is a PhD candidate at the Department of Sociology and Social Research at the University of Trento (Italy). Her main research interests concern educational stratification, inequalities in educational outcomes, teacher bias and the measurement of cognitive and non-cognitive skills, with an analytical approach focusing on mechanisms at the micro-level.

Moris Triventi is an associate professor in the Department of Sociology and Social Research at the University of Trento (Italy); from 2013 to 2016 he was a research fellow at the European University Institute. His research interests comprise social stratification and inequalities, education, crime and policy evaluation. His works have been published, among others, in Annual Review of Sociology, European Sociological Review, Work Employment & Society, and Sociology.

References

Argentin

Triventi

(2011) Social inequality in higher education and labour market in a period of institutional reforms: Italy, 1992–2007. Higher Education 61: 309–323.

Ballarino

Bernardi

Requena

, et al. (2009) Persistent inequalities? Expansion of education and class inequality in Italy and Spain. European Sociological Review 25: 123–138.

Ballarino

Panichella

(2016) Social stratification, secondary school tracking and university enrolment in Italy. Contemporary Social Science 11: 169–182.

Barone

Assirelli

Abbiati

, et al. (2018b) Social origins, relative risk aversion and track choice: A field experiment on the role of information biases. Acta Sociologica 61: 441–459.

Barone

Luijkx

Schizzerotto

(2010) Elogio dei grandi numeri: Il lento declino delle disuguaglianze nelle opportunità di istruzione in italia. Polis (Bologna, Italy) 24: 5–34.

Barone

Triventi

Assirelli

(2018a) Explaining social inequalities in access to university: A test of rational choice mechanisms in Italy. European Sociological Review 34: 554–569.

Barone

Triventi

Facchini

(2021) Social origins, tracking and occupational attainment in contemporary Italy. Longitudinal and Life Course Studies 12: 441–462.

Becker

Mulligan

(1997) The endogenous determination of time preference. The Quarterly Journal of Economics 112(3): 729–758.

Becker

Glauser

(2018) Berufsausbildung, berufsmaturität oder mittelschule? Soziale selektivität beim Übergang in die sekundarstufe II in der deutschschweiz. Swiss Journal of Sociology 44: 9–33.

10.

Becker

Hecken

(2009) Higher education or vocational training? An empirical test of the rational action model of educational choices suggested by breen and goldthorpe and esser. Acta Sociologica 52: 25–45.

11.

Blossfeld

Buchholz

Skopek

, et al. (2016) Models of Secondary Education and Social Inequality. An International Comparison. Cheltenham, UK: Edward Elgar.

12.

Boudon

(1974) Education, Opportunity, and Social Inequality. Changing Prospects in Western Society. Wiley Series in Urban Research. New York: Wiley .

13.

Bourdieu

(1977) Cultural reproduction and social reproduction. In: Karabel

Halsey

(eds) Power and Ideology in Education. New York: Oxford University Press, 487–511.

14.

Bourdieu

(1984) Distinction: A Social Critique of the Judgement of Taste. Cambridge: Harvard University Press.

15.

Bourdieu

Passeron

(1990) Reproduction in Education, Society, Culture. London: Sage.

16.

Breen

Goldthorpe

(1997) Explaining educational differentials: Towards a formal rational action theory. Rationality and Society 9(3): 275–305.

17.

Breen

Luijkx

Müller

, et al. (2009) Nonpersistent inequality in educational attainment: Evidence from eight European countries. American Journal of Sociology 114: 1475–1521.

18.

Breen

van de Werfhorst

Jæger

(2014) Deciding under doubt: A theory of risk aversion, time discounting preferences, and educational decision-making. European Sociological Review 30: 258–270.

19.

Breen

Yaish

(2006) Testing the Breen-Goldthorpe model of educational decision making. In: Morgan

Grusky

Fields

(eds) Mobility and Inequality. Frontiers of Research From Sociology and Economics. Stanford, CA: Stanford University Press, 232–258.

20.

Checchi

Flabbi

(2013) Intergenerational mobility and schooling decisions in Germany and Italy: The impact of secondary school tracks. Rivista di Politica Economica 3: 7–57.

21.

Cobalti

Schizzerotto

(1994) La Mobilità Sociale in Italia. Bologna: Il Mulino.

22.

Contini

Scagni

(2013) Social-origin inequalities in educational careers in Italy: Performance or decision effects? In: Jackson

(eds) Determined to Succeed? Performance Versus Choice in Educational Attainment. Stanford: Stanford University Press, 149–184.

23.

Contini

Triventi

(2016) Between formal openness and stratification in secondary education: Implications for social inequalities in Italy. In: Blossfeld

Buchholz

Slopek

Triventi

(eds) Models of Secondary Education and Social Inequality: An International Comparison. Cheltenham, UK: Edward Elgar, 305–322.

24.

Elster

(1979) Ulysses and the Sirens. Cambridge: Cambridge University Press.

25.

Erikson

Jonsson

(1996) Can Education be Equalized? The Swedish Case in Comparative Perspective. Boulder, Colo: Westview Press.

26.

Gabay-Egozi

Shavit

Yaish

(2010) Curricular choice: A test of rational choice model of education. European Sociological Review 26: 447–463.

27.

Gambetta

(1987) Studies in Rationality and Social Change. Were They Pushed or Did They Jump?: Individual Decision Mechanisms in Education. Cambridge: Cambridge University Press.

28.

Guetto

Vergolini

(2017) Educational expansion without equalization: A reappraisal of the ‘effectively maintained inequality’ hypothesis in children's choice of the upper secondary track. European Societies 19: 1–27.

29.

Hillmert

Jacob

(2003) Social inequality in higher education: Is vocational training a pathway leading to or away from university? European Sociological Review 19(3): 319–334.

30.

Holm

Jæger

(2008) Does relative risk aversion explain educational inequality? A dynamic choice approach. Research in Social Stratification and Mobility 26: 199–219.

31.

Jackson

(2013) Determined to Succeed? Performance Versus Choice in Educational Attainment. Stanford: Stanford University Press.

32.

Jann

(2014) Plotting regression coefficients and other estimates. The Stata Journal 14: 708–737.

33.

Jæger

Breen

(2016) A dynamic model of cultural reproduction. American Journal of Sociology 121: 1079–1115.

34.

Karlson

Holm

(2011) Decomposing primary and secondary effects: A new decomposition method. Research in Social Stratification and Mobility 29(2): 221–237.

35.

Kohler

Karlson

Holm

(2011) Comparing coefficients of nested nonlinear probability models. The Stata Journal 11(3): 420–438.

36.

Lucas

(2001) Effectively maintained inequality: Education transitions, track mobility, and social background effects. American Journal of Sociology 106: 1642–1690.

37.

Müller

Karle

(1993) Social selection in educational systems in Europe. European Sociological Review 9: 1–23.

38.

Obermeier

Schneider

(2015) Educational choice and risk preferences: How important is relative vs. Individual risk preference? Journal for Educational Research Online 7: 99–12.

39.

Panichella

Triventi

(2014) Social inequalities in the choice of secondary school. European Societies 16: 666–693.

40.

Ress

Azzolini

(2014) Primary and secondary effects of social background on educational attainment in Italy. Evidence from an administrative dataset. Italian Journal of Sociology of Education 6: 53–80.

41.

Schizzerotto

(1997) Perché in italia ci sono pochi diplomati e pochi laureati? Vincoli strutturali e decisioni razionali degli attori come cause della contenuta espansione della scolarità superiore. Polis, Ricerche e Studi su Società e Politica 3: 345–366.

42.

Simon

(1955) A behavioral model of rational choice. Quarterly Journal of Economics 69: 99–118.

43.

Stockè

(2007) Explaining educational decision and effects of families’ social class position: An empirical test of the Breen-Goldthorpe model of educational attainment. European Sociological Review 23(4): 505–519.

44.

Triventi

Barone

Facchini

(2021) Upper secondary tracks and student competencies: a selection or a causal effect? Evidence from the Italian case. Research in Social Stratification and Mobility. https://www-sciencedirect-com-443.web.bisu.edu.cn/science/article/abs/pii/S0276562421000469.

45.

Triventi

Panichella

Ballarino

, et al. (2016) Education as a positional good: Implications for social inequalities in educational attainment in Italy. Research in Social Stratification and Mobility 43: 39–52.

46.

Triventi

Skopek

Kulic

, et al. (2019) Advantage ‘finds Its way’: How privileged families exploit opportunities in different systems of secondary education. Sociology 54: 237–257.

47.

Triventi

Trivellato

(2009) Participation, performance and inequality in Italian higher education in the 20th century. Evidence from the Italian longitudinal household survey. Higher Education 57: 681–702.

48.

van de Werfhorst

Hofstede

(2007) Cultural capital or relative risk aversion? Two mechanisms for educational inequality compared. The British Journal of Sociology 58: 391–415.

49.

Zimmermann

(2020) Social influence or rational choice? Two models and their contribution to explaining class differentials in student educational aspirations. European Sociological Review 36: 65–81.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.44 MB