Hedging Risk in Curricular Choice: A Test of a Rational Choice Model of Education* Manuscript's total word count (excluded tables): 6,934 Limor Gabay-Egozi, Yossi Shavit Tel Aviv University and Meir Yaish University of Haifa Prepared for presentation at the Session on Social Stratification, at the 2007 Annual Meeting of the American Sociological Association, New York City, and at the Montreal Meeting of RC28, 2007 * This study was supported by an Eshkol Fellowship from the Israeli Ministry of Science and Technology, and by a fellowship from the Horowitz Institute at Tel Aviv University to the first author. Please direct correspondence to [email protected] 1 Abstract Rational choice theory of education views student's educational decision as a sequence of binary choices between options that entail long-term utility and options that reduce short-term risk of failure. The model asserts that choice between options is affected by students' utility considerations, failure expectations in either option, and motivation to avoid downward social mobility. We evaluated these assumptions using data on students' choices of school majors in Tel Aviv high schools and found that inequalities between social strata in educational choice were not mediated by the mechanisms proposed by the model. Moreover, and importantly, many students (the hedgers) did not choose between long-term utility and short-term risks but combined the two. Hedgers combined school subjects that are expected to yield long-term utility, such as the hard sciences, with those that reduce the risk of failure in the short term, such as social sciences. The hedgers proved to be drawn disproportionately from disadvantaged classes and were more likely to be women. These results suggest that educational systems that allow multiple rather than binary choices enhance the attainment of working-class youth by enabling them to combine risky options, which may enhance their mobility, with safer choices, which assure success in the short term. 2 Introduction In recent years policy makers have advocated a degree of parental choice in education (Hirsch 1994, Hayman et al. 1997, Plank and Sykes 2003), suggesting that choice will enhance equality of opportunity. Aiming to equalize educational opportunities, most countries have made wideranging changes in education systems, including choice policy such as de-tracking, increased parental choice of school and the introduction of voucher systems. Nevertheless, as argued by some, choice could be a stratifying rather than an equalizing mechanism in the educational attainment process. For example, Lucas argued that despite the replacement of overt tracking by apparent choice in course selection, de facto tracking persisted as did class and race-based inequality therein (Lucas 1999). Ayalon studied curricular choice in Israeli high schools which differ in the degree of choice available to students. She found that gender and socioeconomic differences were more pronounced where choice was abundant (Ayalon 2006). The implications of educational choice for educational stratification therefore merit study. One of the main debates in the literature on educational stratification is encapsulated in the title of Gambetta’s (1987) book Were They Pushed or Did They Jump? Push factors refer to the various constraints that determine students' educational attainment while pull factors refer to choice. These concepts echo Boudon's (1974) distinction between primary and secondary effects. Primary effects are factors responsible for the association between social origins and children’s academic ability and performance, while secondary effects refer to factors that account for educational choices. In recent decades interest has grown in rational-choice explanations of educational stratification (e.g., Boudon 1974, Gambetta 1987, Goldthorpe 1996, 1998, Erikson and Jonsson 1996, Breen and Goldthorpe 1997, Morgan 1998, 2005). Rational-choice theory assumes that 3 individuals are not only governed by push factors operating behind their backs, but are conscious decision makers whose actions are influenced by a costs-benefit calculus (Hedström and Stern forthcoming). This paper tests three important assumptions of the Rational Choice Model of education. Theoretical Considerations Studies which focus on push factors concentrate on two sets of variables affecting educational stratification: familial resources and characteristics of the educational system. Among them are families' economic resources (Duncan et al. 1998), cultural resources (De Graaf et al. 2000), students’ track placement (Shavit 1984, 1990) and the curriculum offered in schools (Apple 1990, Oakes 1990, Ayalon 1995, 2002, Burkam et al.1997). Choice, rather than constraint, has lately attracted increasing attention as a stratifying mechanism in the educational attainment process. Scholars interested in pull factors argue that inequality is also due to processes involving choice (e.g., Goldthorpe 1996, Breen and Goldthorpe 1997). Several studies have shown that class differentials in educational attainment persist even after controlling for push factors and for scholastic performance (cf. Erikson and Jonsson 1996, Breen and Jonsson 2000, Need and de Jong 2001, Davies et al. 2002, Breen and Yaish 2006). In England and Wales, for example, class differentials in educational choice account for about 25 percent of inter-class educational inequality (Erikson et al. 2005, Jackson et al. forthcoming). The growing interest in secondary factors as an educational stratification mechanism has led scholars to develop theoretical models using broader notions of rationality, aiming to explain why children of similar abilities but different class backgrounds are observed to make different 4 educational choices (Goldthorpe 1996, 1998, Erikson and Jonsson 1996, Breen and Goldthorpe 1997, Morgan 1998, 2005). Breen and Goldthorpe's (1997) Rational Action Theory is arguably the most influential of these models. These models share certain properties. First, the educational attainment process is viewed as a sequence of transition points at which students (and their families) decide whether to drop out or to continue in any of the available educational options to the next level. Each available option is evaluated and ranked on the basis of its utility and risk. Utility is defined as the degree to which it enhances occupational and economic attainment. Risk is defined as the odds of failing to complete a prescribed course of study successfully due to the perceived difficulty and the effort required. It is assumed that options associated with high utility are also associated with high risk of failure, so decision makers face an inherent dilemma between maximizing the former and minimizing the latter. To anticipate, our own results reveal that many students hedge short-term risks and long-term utility. They choose both high-utility and low-risk options, hoping thereby to reap the long-term benefits of the former while insuring themselves against short-term failure. Hedging is possible in educational systems that allow multiple rather than binary choices. Second, although students from different social classes rank all options equally in terms of their associated utility and risk, their own subjective probability of success is formed on the basis of past performance at school. Finally, Breen and Goldthorpe's (1997) Relative Risk Aversion mechanism assumes that students from all social classes alike aim to maintain their parental class position and choose educational options which, they believe, will minimize the risk of downward mobility (Breen and Goldthorpe 1997: 283-5). Since students of all origins wish to maintain their class position, the implications of the relative risk aversion mechanism differ by class background: for middle-class 5 children it implies choosing the risky option because this alone will lead them to the upper classes. Conversely, working-class children tend to choose less risky options because they are content with attaining working-class occupations.1 In this paper we test these assumptions. More specifically, we test the following three hypotheses: Hypothesis I: Students' educational choices are affected by their beliefs about the likely returns to the various educational credentials. Hypothesis II: Young students' educational choices are driven by their beliefs about their own odds of success or failure in the alternative educational trajectories. Hypothesis III: Educational decisions are motivated by individuals' apprehension of downward class mobility (i.e., the relative risk aversion motivation). The Setting We tested these assumptions with data on the choice of major subjects by secondary school students in Tel Aviv, Israel. Our sample consisted of ninth and tenth graders who were about to choose a major for their matriculation examinations. The choice of advanced courses and examinations is an important junction in the socioeconomic life course of young Israelis. At age twelve, after a year in pre-school and six years in primary school, Israeli children enter middle schools where they spend grades seven, eight and nine. This spell is followed by upper secondary school in grades ten through twelve, where students are prepared for the matriculation 1 When weighing the pros and cons of each educational option, students (and parents) are assumed to take into account, among other things, the direct and indirect costs of education. Although perceived cost is an important variable in rational-choice models of education, it is not measured in this study. In Israel, high school majors hardly differ in their expected costs, so this element is disregarded in both the theoretical and empirical parts of the paper. 6 examinations leading to the award of the matriculation certificate (bagrut), which is required for higher education.2 The matriculation exams are offered on a unit basis: they can be taken at the basic level (3 units) or the advanced levels (4 or 5 units).3 To earn the certificate students must take exams in subjects totaling at least 20 units. Seven subjects are compulsory, two of them being English and mathematics. They must also take an elective major at the advanced level. Research shows that students from advantaged class origins tend to specialize in sciences, while those from lower classes and ethnic groups tend to specialize in the humanities and the social sciences (Ayalon 1994, 1995, Ayalon and Yogev 1997). The major is usually chosen during the spring term of tenth grade (although in one of the schools we studied students made the choice as early as ninth grade). Admission into higher education predicates a matriculation certificate, and the more prestigious departments and institutions of higher education set several additional requirements: (1) advanced English (at least 4 units); (2) a high matriculation grade point average; and (3) some departments specifically require applicants to have passed advanced matriculation exams in math and sciences. Moreover, the complicated formula employed by universities tends to favor applicants who have passed more than one advanced level examination. In recent cohorts, over 70 percent sat for one matriculation examination or more, but only half of those were eligible for the certificate (Ayalon and Shavit 2004). Math and English are the 2 Upper secondary education consists of two main tracks: academic and vocational. Until recently the vocational track trained most students in a vocation and prepared them for the world of work rather than for further study. In recent years, however, most vocational-track students have been able to sit for matriculation examinations, although their success rates are lower than those of their academic counterparts (e.g., Ayalon and Shavit 2004). 3 The number of units refers to the time devoted to the subject (one unit is equal to one hour a week for three years, or three hours a week for one year), which corresponds to the subject's level and degree of difficulty (Ayalon, 1994) . 7 major stumbling blocks: only about 35 and 40 percent, respectively, of recent birth cohorts passed them, compared, for example, with 50 percent who passed Hebrew language. In an attempt to raise students' success rates in math and English, schools actively determine their unit level as early as ninth grade. Once the level has been determined, a student is tracked and has little further choice in this regard. Data In the spring term of 2006 we distributed self-administered questionnaires among tenth graders attending four Hebrew secular, non-vocational public schools in Tel Aviv-Jaffa.4 At one school we administered the questionnaire to ninth graders also. Tel Aviv-Jaffa consists of nine administrative districts which are fairly homogeneous socio-economically (Tel Aviv-Jaffa 2006). Tel Aviv-Jaffa has nine secular, non-vocational secondary schools, of which two are atypical and were not included in our target population: one specializes in the arts and the other caters primarily to Russian immigrant children. Of the remaining seven schools we sampled two in working-class districts and two in middle-class districts. Fortunately, all four school principals cooperated and gave us access to their students. We handed out the questionnaires in 28 4 We restricted our sample to academic and secular high schools because the three sectors – academic, vocational and religious schools – differ in the curricular choices on offer and in the relative utility and prestige that students attribute to subjects. Vocational schools offer a more limited menu of academic subjects along with an abundance of technical ones. In religious schools strong students tend to take religious studies rather than sciences, as is the case in secular schools. For simplicity, we preferred to limit the study to secular academic schools, where curricular options are relatively similar. 8 classrooms and obtained 683 completed forms, of which 17 percent were excluded from the analysis due to missing values.5 The Dependent Variable Our dependent variable is a three-category classification representing choice of advanced subjects. The categories are: hard subjects (physics, chemistry and computer science), soft subjects (biology, economics, social sciences, history, literature and communication), and a mixture of both. Next we explain and justify the classification. Respondents were presented with a list of subjects available at their school and were asked to indicate whether or not they planned to take each subject at an advanced level. The distribution of choices is shown in Table 1.6 As seen, boys tended to choose physics, computers, economics and other social sciences, while girls tended to choose the social sciences, communications, economics and biology. [TABLE 1 HERE] From a rational-choice perspective, the subjects should be distinguished on the basis of students' beliefs about their utility and according to students' perception of the risk of failure associated with each subject. We measured the former by asking respondents the following 5 Students' sampling probabilities were computed as the product of the school's sampling productivity (0.29 and 0.66 in the middle class and working class neighborhoods, respectively) and the students' sampling probability within the school, which was computed as the ratio of the number of respondents and the number of tenth graders in the school (ninth and tenth graders in one of the schools). In the analysis the cases were weighted inversely to their sampling probabilities. 6 Recall that students' placement in English and math is largely determined by teachers, and that students have little choice in the matter. Therefore, we did not count these subjects among the available options from which curricular choice is made. 9 questions: "In your opinion, if a student succeeds in this subject at the 5-unit level, what are his or her chances of admission to a university?" [scale: 1 ('not high at all') to 5 ('very good')]. The mean responses for each subject are shown in column 1 of Table 2. Students perceived physics, chemistry, and computers to be of most utility with respect to university admission, while the social sciences, history, literature and communications were perceived to be of less utility. Economics and biology occupied intermediate positions. [TABLE 2 HERE] Students' perception of risk was measured as the proportion who thought that their grade in a subject would be low (64 or less on a 100-point scale) if they took it. These proportions are shown in column 2 of Table 2. Physics was perceived as the most risky, followed by chemistry and computers. The social sciences, history, literature and communications were perceived as the least risky, while economics and biology occupied intermediate positions. Thus, the hard subjects – physics, chemistry and computers – were perceived to be of most utility but also the most risky. We refer to all other subjects, including economics and biology, as soft subjects. [TABLE 3 HERE] As indicated earlier, although for a matriculation certificate students are required to take just one subject at an advanced level, incentives exist for them to take more. In fact, 96 percent of the students in our sample intended to take two or more advanced subjects. As seen in Table 3, 21 percent of the sample intended to take advanced subjects in the hard sciences alone, 52 percent intended to take soft subjects only and 24 percent intended to take a mix. Thus, over half the students who intended to take a hard subject seem to hedge their risk of failure by taking a 10 soft one too. This strategy seems quite reasonable given that students are only required to complete a single advanced course and can discard an extra course in which they might do poorly. In the rest of the analysis we therefore employ a multiple choice model which distinguishes three choices: risky but rewarding hard subjects, less risky and less rewarding soft subjects, and a mix which hedges risk. Our expectation is that, net of scholastic performance, any remaining class differences in the choice among these alternatives will be explained by secondary factors. Secondary Factors The rational choice model of education refers to three secondary factors affecting educational choice: relative risk aversion, beliefs about future utilities, and failure expectation. Relative risk aversion was measured by the following battery of questions: "To what extent do you agree with each of the following statements?" [scale: 1 ('strongly disagree') to 5 ('strongly agree')] a. It is important for me that my salary be at least equal to the level of my parents' salary. b. My parents would not be satisfied if I were to work at a lower occupation than theirs. c. I would like to attain a social class on a level at least equal to my parents'. We factor-analyzed these items and found them to load on a single factor, which we interpret to represent the class maintenance motivation (i.e., relative risk aversion). Beliefs about future utilities was measured by the question shown earlier about the subjective odds that a student who succeeds in each subject will enter university. For each student we computed the mean perception of success associated with the hard subjects and the 11 soft subjects. The ratio of the former to the latter was then our measure of the relative utility that students attribute to the hard subjects as against the soft subjects. Relative Failure Expectation was measured by students' assessment of their expected performance in each subject were they to take it at an advanced level. Within each cluster of pure subjects – the hard and the soft – we calculated for each student the proportion of those in which he or she expected to earn a low grade (64 or less). We then computed the difference between the former and latter proportion. High values represent apprehension of the hard subjects. Primary Factors, Scholastic Performance and Grouping Two variables indicate social origins: parental education and family's economic resources. Parental education is measured as the highest qualification obtained by either father or mother. It is measured on a six-category classification which maintains a clear hierarchical order. Students were also asked to provide information on the availability of a variety of durable goods in the home (such as air-conditioning, computer, dishwasher, car etc.). Economic resources are measured as the sum of available items in the parental household.7 Students' scholastic performance is indicated by their grades, measured as the mean selfreported grade in Hebrew, English and math on the most recent report card. Grouping is the level at which students considered themselves placed in tenth-grade math. Originally grouping was measured on a three-point scale corresponding to levels 3 (regular), 4, and 5 (advanced). In the following analyses grouping is measured as a dummy variable indicating advanced level. As noted earlier, teachers assign students to math levels, 7 In unreported analyses we also controlled for father's occupational SEI, number of siblings, and students' reading habits, but most of their effects did not reach statistical significance. Dropping them from the analysis did not alter the substance of our results in any way. 12 usually in ninth grade. Group placement is largely determined by students’ prior performance in math, but once placement has been made group is an evident push factor. It reflects students' earlier achievements, but it also shapes their self-esteem and confidence in math and constrains their future choices. Students placed at regular-level math are unlikely to choose advanced hard subjects. Finally, we also control for sex, with a dummy variable indicating boys. Having made these preliminary qualifications we can embark on the analysis. Analysis Table 4 presents descriptive statistics of the independent variables by our three categories of advanced matriculation subjects. The results in Table 4 indicate significant gender differences in the choice of matriculation subjects. Boys are over-represented in the hard subjects and in the mixed category (which includes at least one hard subject), and are under-represented in the soft subjects. This gender difference in the selection of school majors at high school, and later in higher education, is consistent with previous studies in Israel (Ayalon and Yogev 1997, Ayalon 2003, Katz-Gerro and Yaish, 2003) and elsewhere (Ma and Willms 1999, Bradley 2000). With regard to primary factors, Table 4 shows, as expected, that students who choose the more demanding, yet more rewarding, matriculation subjects (i.e., hard subjects) are from a more advantaged social background than those who choose the less demanding subjects (cf. Oakes et al. 1992; for Israel see Ayalon 1994, 1995, Ayalon and Yogev 1997). However, students who choose the least demanding matriculation subjects (i.e., soft subjects) are found to be from a relatively more advanced social background than those who select a combination of hard and soft subjects. This is an interesting result that we shall return to later. Regarding scholastic performance and grouping, the entries in Table 4 reveal that students who choose hard subjects display academic excellence, while those who select the 13 mixed or the soft subjects tend to achieve lower grades. Similarly, those who choose hard subjects are more likely to have been placed in advanced grouping in math. [TABLE 4 HERE] As for secondary factors, Table 4 shows that students who choose mixed subjects exhibit the highest degree of relative risk aversion. Students who are highly concerned with class maintenance apparently hedge their risk by adopting a mixed strategy. The hard subjects increase the expected long-term utility of the mix (i.e., higher education and labor market utilities), while the soft subjects cushion the consequence of possible failure in the short term (i.e., failing to obtain a matriculation certificate). If one does poorly in the hard (and risky) subjects, success in the soft ones averts short-term failure. Choosing a mix of subjects can be seen as a hedging strategy designed to minimize exposure to unwanted short-term risks, while allowing the possibility of long-term success in higher education and in the labor market. Finally, and as expected, those who select hard subjects hold stronger beliefs about the positive utility of the hard subjects than those who choose the other combinations of subjects; those who select soft subjects have the lowest confidence in their ability to succeed in hard subjects. Primary Effects on Academic Performance, Grouping and Secondary Factors Before testing our three hypotheses, we estimate in Table 5 the effect of social background on performance and grouping, and that of social background, performance and grouping on secondary factors. 14 Consistent with previous research, social background has positive effects on both academic performance and grouping. Further, boys and girls are seen not to differ in their academic performance, though more girls than boys are placed in lower grouping levels in math. In the last three columns of Table 5 we present the results of OLS regression models on secondary factors: relative risk aversion, beliefs about future utilities, and failure expectation. Table 5 indicates that students of a more advantaged social background are less concerned with maintaining their parental socioeconomic positions, and tend to attribute higher future utility to the hard rather than the soft subjects. A plausible explanation for the negative association between relative risk aversion and social background might be related to the degree of confidence that people of different social backgrounds have in their social position. Students of advantaged social origins may take their position for granted, and are therefore less concerned with status maintenance. By contrast, students from less privileged families may suffer from status anxiety and fear of status loss. [TABLE 5 HERE] Referring to beliefs about utility, the effect of social origin on relative utility expectation should be seen in the context of class-based inequality in information, whereby these beliefs are formed. Studies have shown large class differences in the amount and the accuracy of information acquired about education: the working class proves less informed than the advantaged classes (Hayman 2001, Ball et al. 1995, Lareau 1987, Schneider et al. 2000). Less informed people, moreover, are less discriminating among educational alternatives (Morgan, 2005).8 8 Table 5 also indicates gender differences in secondary effect. Although this issue is of considerable importance it is beyond the scope of this paper. We hope to treat it in a separate study. 15 Primary and Secondary Effects on Choice of Advanced Subjects Testing the three hypotheses, we ask: are students' educational choices affected by their apprehension of downward class mobility (i.e., relative risk aversion), by their beliefs about the likely returns on the various choices, and by their beliefs about their own odds of success or failure in the alternative trajectories? We also ask: do these secondary factors explain away any remaining class inequalities in educational choice? That is, is choice a stratifying mechanism? To answer these questions we estimate a multinomial logit model on three curricular choice clusters, consisting of (a) pure hard subjects, (b) pure soft subjects, and (c) a mix. The model includes the following independent variables: sex, parental education, family's economic resources, scholastic performance, grouping in math, relative risk aversion, subjective utility of the hard as against the soft subjects, and failure expectation in the hard as against the soft subjects. Tables 6a, 6b and 6c present the parameter estimates of each pair of contrasts among these choices: hard subjects vis-à-vis soft subjects (Table 6a), hard subjects vis-à-vis a mix (Table 6b), and soft subjects vis-à-vis a mix (Table 6c). For ease of comparability in and between the models all independent variables, except the sex and grouping dummies, were standardized to a mean of zero and standard deviation of one. [TABLE 6a HERE] Hard Subjects vis-à-vis Soft Subjects Model I in Table 6a indicates that males, and students with educated parents, are more likely to opt for hard than for soft subjects. When we add academic performance to this model (Model II) the effects of parental education and economic resources are sharply reduced, and adding grouping (Model III) renders the effect of parental education even weaker and statistically 16 insignificant. The implication of this result is that choice between the two homogeneous clusters of subjects is largely predetermined by performance and grouping, and that no effects of social background remain to be explained by secondary factors. In Model IV the analysis focuses on the gross effects of the three secondary factors. The relative risk aversion mechanism proves not to exert its expected effect on choice between hard and soft subjects. By contrast, the other two secondary factors – relative utility and relative failure expectation – are significant, and in the expected direction: students who expect relatively higher utility from the hard subjects and hold low failure expectations in them are more likely to select hard than soft subjects. When, in Model VI, we also control for social background, academic performance and grouping the net effects of the secondary factors remain largely unchanged, compared with Model IV. In sum, the main determinants of this specific educational choice are: academic performance, grouping, relative failure expectation and subjective utility, but not relative risk aversion. Secondary factors do not seem to mediate the effects of social background. Hard Subjects vis-à-vis a Mix of Subjects In Table 6b we see again, as expected, that parental education and economic resources are positively associated with the choice of hard subjects rather than a mix of subjects. Interestingly, net of academic performance (Model II) and grouping (Model III), social background still exerts statistically significant effects on this educational choice. Do secondary factors explain these remaining effects of background? [TABLE 6b HERE] 17 Before answering this question we examine, in Model IV, the gross effects of secondary factors on this specific choice. Expectedly, students who attribute relatively high utility to the hard subjects tend to choose them. More interestingly, students with strong motivations for status maintenance are seen to be less likely to select a pure cluster of hard subjects and seem to prefer a mix. This suggests that those who are strongly motivated to maintain their family class position tend to hedge risk by choosing a mix of subjects. Hedging reduces the risk of failure in the short term while leaving open the option of benefiting from the long-term utility afforded by the hard subjects. We know from Table 5 that students from relatively less affluent and less educated families are more concerned with class maintenance than those from more affluent and more educated families. This implies that hedging in this context is associated with members of the less privileged strata in society. This assertion is corroborated by the results of Model VI, which show positive effects of background on the choice of hard subjects rather than a mix. Finally, a comparison of Models III and VI shows that secondary factors do not mediate the effects of social background on this particular choice. Soft Subjects vis-à-vis a Mix of Subjects As can be seen in Model I of Table 6c, girls and students from privileged social strata are more likely to choose soft subjects than a mix. In Model III we observe that performance does not affect the choice under investigation, while grouping in math is positively associated with a mix of subjects. This latter result is not surprising as the mix category includes, by definition, hard subjects, which normally require advanced math. We see further that the effect of parental education is larger in Model III than in Model I, such that net of performance and grouping, 18 students from advantaged families are more likely to choose soft subjects than a mix.9 Again, students of less advanced origins are more likely to opt for the mixed strategy. This result is consistent with our claim that choosing the mixed category represents hedging behavior. [TABLE 6c HERE] We would argue that among equally able students, the less privileged classes are more wary of putting all their eggs in the soft-subject basket – these generally being perceived as less rewarding than the hard subjects – for fear of failing to win admittance to university. Therefore, they hedge this long-term risk by taking hard subjects as well, which they believe will enhance their long-term objectives. The privileged classes, on the other hand, are less apprehensive about their long-term life chances. They are confident in the help that their family can provide, and are less concerned with hedging behavior in curricular choice. The effects of social background hardly change when secondary factors are added to the model (Model VI). That is, secondary factors do not seem to mediate the effects of social origins. Summary and Discussion It is generally recognized that students of a privileged social background progress farther in the educational system than those of less advantageous origins. Rational choice models of education attribute these inequalities to differences in educational choice made by individuals from different social strata. This paper set out to test the main propositions of these models. 9 This change is due to the suppressor effect of Grades and Grouping on the association between parental education and choice: parental education affects performance and grouping positively (see Table 4), and these in turn reduce the odds of choosing soft subjects. 19 According to these models, the educational process consists of a sequence of branching points at which students are faced with binary choices. On the one hand, the high road is demanding and risky in the short term, but can lead to rewarding life-course outcomes in the long run. On the other hand, the low road is less demanding and safer in the short term, but can offer only limited long-term rewards. That is, long-term utility and short-term risks are positively associated. It is further assumed that on reaching educational decision points students consider their success probability in each alternative; its utility, defined as the probability that it will lead to desirable occupational outcomes; and the extent to which it will protect them against downward social mobility. According to Breen and Goldthorpe's (1997) version of the model, the desire to avoid downward mobility gives rise to class differentials in educational choice. Students from the upper strata are impelled to choose the high road, which is necessary if they are to maintain their social position, while those of lower strata choose the low road, which is sufficient to maintain their social position. Finally, rational-choice models postulate that social reproduction is mediated, in part, by differential educational choice. Here we put these assumptions to empirical test. More to the point, we examined the role of utility considerations, of failure expectations in education, and of class maintenance motivation in shaping educational choice. We examined the validity of these assumptions by analyzing recently collected data from a purpose-designed survey of Tel Aviv-Jaffa high school students who in 2006 were about to select school subjects at an advanced level for their matriculation examinations. In the main, our results cast doubt on the validity of the rational choice assumptions, and indeed on the entire model. We now deal with each of these assumptions in turn. Although we found that educational choice was affected by class maintenance motivation, as well as by subjective utility 20 and failure expectations, these effects did not mediate class inequalities in educational choice, which were largely mediated by performance and prior grouping. Moreover, and importantly, a large proportion of our respondents took neither the high road nor the low road, preferring a mixture of the two. We argue here that students who choose high-utility options and low-risk ones hedge short-term risks and long-term utility. We find that hedgers came disproportionately from less affluent and less educated families, had lower academic achievements, and were more concerned with class maintenance, than those who chose either pure hard subjects or pure soft subjects. The possibility to choose a mix of subjects is not anticipated by rational choice models of education. This is an important finding indicating that the model is perhaps incomplete. It ignores systemic differences in the structure of available choices. The availability of multiple rather than binary options can affect educational choice and class-based educational inequality by allowing students to hedge long-term utility with short-term risk. Comparative research across educational systems that present students with a variety of alternatives to choose from is in order. Our finding that the less privileged in society adopted a hedging strategy also calls into question the argument that curricular choice masks transparent tracks visible only to the more affluent parents (Lucas 1999). We find that students from less affluent and less educated families seemed to be aware of the risks and utilities associated with the different subjects available to them, and they mixed them to their apparent advantage. The availability of options need not work to the detriment of greater equality of educational opportunity, as suggested by Lucas (also by Ayalon 2006). Research is also called for on the relationship between students' stated intentions and the courses they actually take. While actual course taking is affected by students' preferences, it is also 21 affected by the availability of courses at schools, by teachers' recommendations, by schools' policies of student selection, and by a host of other institutional constraints. In a future study we hope to evaluate the role of subjective choices and of hedging, as well as of push factors, on the actual courses students take. 22 REFERENCES Apple, Michael W. (1990). Ideology and Curriculum (2nd edition). New York: Routledge. Ayalon, Hanna (1994). "Monopolizing Knowledge - the Ethnic-Composition and Curriculum of Israeli High-Schools." Sociology of Education 67(4): 264-278. Ayalon, Hanna (1995). "Math as a Gatekeeper: Ethnic and Gender Inequality in Course Taking of the Sciences in Israel." American Journal of Education 104(1): 34-56. Ayalon, Hanna (2002). "Mathematics and Sciences Course Taking among Arab Students in Israel: A Case of Unexpected Gender Equality." Educational Evaluation and Policy Analysis 24(1): 63-80. Ayalon, Hanna (2003). "Women and Men Go to University: Mathematical Background and Gender Differences in Choice of Field in Higher Education." Sex Roles 48(56): 277-290. Ayalon, Hanna (2006). "Nonhierarchical Curriculum Differentiation and Inequality in Achievement: A Different Story or More of the Same?" Teachers College Record 108(6): 11861213. Ayalon, Hanna and Adam Gamoran (2000). "Stratification in Academic Secondary Programs and Educational Inequality in Israel and the United States." Comparative Education Review 44(1): 54-80. Ayalon, Hanna and Yossi Shavit (2004). "Educational Reforms and Inequality in Israel: The MMI Hypothesis Revisited." Sociology of Education 77(2): 103-120. 23 Ayalon, Hanna and Abraham Yogev (1997). "Students, Schools, and Enrollment in Science and Humanity Courses in Israeli Secondary Education." Educational Evaluation and Policy Analysis 19(4): 339-353. Ball, Stephen J., Richard Bowe and Sharon Sewirtz (1995). "Circuits of Schooling: A Sociological Exploration of School in Social Class Context." Sociological Review 43(1): 52-78. Boudon, Raymond (1974). Education, Opportunity, and Social Inequality: Changing Prospects in Western Society. New York: Wiley. Bradley, Karen (2000). "The Incorporation of Women into Higher Education: Paradoxical Outcomes?" Sociology of Education 73(1): 1-18. Breen, Richard and John H. Goldthorpe (1997). "Explaining Educational Differentials - Towards a Formal Rational Action Theory." Rationality and Society 9(3): 275-305. Breen, Richard and Jan O. Jonsson (2000). "Analyzing Educational Careers: A Multinomial Transition Model." American Sociological Review 65(5): 754-772. Breen, Richard and Meir Yaish (2006). "Testing the Breen-Goldthorpe Model of Educational Decision Making." Pp. 323-358 in Mobility and Inequality: Frontiers of Research from Sociology and Economics edited by L. Stephen, David B. Grusky, and Gary. S. Fields. Stanford: Stanford University Press. 24 Burkam, David T., Valerie E. Lee and Becky A. Smerdon (1997). "Gender and Science Learning Early in High School: Subject Matter and Laboratory Experience." American Educational Research Journal 34 (2):297–331. Davies, Richard, Eskil Heinesen and Anders Holm (2002). "The Relative Risk Aversion Hypothesis." Journal of Population Economics 15(4): 683-713. De Graaf, Nan Dirk, Paul M. De Graaf and Gerbert Kraaykamp, G. (2000). “Parental Cultural Capital and Educational Attainment in the Netherlands: A Refinement of the Cultural Capital Perspective.” Sociology of Education 73(2): 92-11. Duncan, Greg J., Jean W. Yeung, Jeanne Brooks-Gunn and Judith R. Smith (1998). "How Much Does Childhood Poverty Affect the Life Chances of Children?" American Sociological Review 63(3): 406-423. Erikson, Robert E., John H. Goldthorpe, Michelle Jackson, Meir Yaish and David R. Cox (2005). "On Class Differentials in Educational Attainment." Proceedings of the National Academy of Sciences of the United States of America 102(27): 9730-9733. Erikson, Robert E. and Jan. O. Jonsson (1996). Can Education Be Equalized? The Swedish Case in Comparative Perspective. Boulder, CO: Westview. Gambetta, Diego (1987). Were They Pushed or Did They Jump? Individual Decision Mechanisms in Education. Cambridge, Cambridge University Press. 25 Goldthorpe, John H. (1996). "Class Analysis and the Reorientation of Class Theory: The Case of Persisting Differential in Educational Attainment." British Journal of Sociology 47(5): 481-505. Goldthorpe, John H. (1998). "Rational Action Theory for Sociology." British Journal of Sociology 49(2): 167-192. Hayman, Flor (2001). The Dynamic of School Choice by Parents and Students. Unpublished Ph.D. dissertation, Tel Aviv University (in Hebrew). Hayman, Flor, Haya Golan and Rina Shapira (1997). "School Autonomy and Parental Choice: Steps in Local Educational Planning." Pp. 77-108 in Autonomy and Change in Context: An International Perspective edited by Rina Shapira and P.W. Cookson. Oxford: Pergamon. Hedström, Peter and Charlotta Stern (forthcoming). "Rational Choice and Sociology." Pp. in The New Palgrave Dictionary of Economics edited by Stephen Durlauf and Lawrence Blume (2nd edition). New-York: Palgrave Macmillan. Hirsch, Morris. W. (1994). "Saturation at High Gain in Discrete Time Recurrent Networks." Neural Networks 7(3): 449-453. Jackson, Michelle, Robert E. Erikson, John H. Goldthorpe and Meir Yaish (forthcoming). "Primary and Secondary Effects in Class Differentials in Educational Attainment: the Transition to A-Level Courses in England and Wales". Acta Sociologica. 26 Katz-Gerro, Tali and Meir Yaish (2003). "Higher Education: Is More Better? Gender Differences in Labour Market Returns to Tertiary Education in Israel." Oxford Review of Education 30(1): 571-592. Lareau, Annette (1987). "Social Class Differences in Family-School Relationships: The Importance of Cultural Capital." Sociology of Education 60(2): 73-85. Lucas, Samuel R. (1999). Tracking Inequality: Stratification and Mobility in American High Schools. New York: Teachers College Press. Ma, Xin and Douglas. J. Willms (1999). "Dropping Out of Advanced Mathematics: How Much Do Students and School Contribute to the Problem?" Educational Evaluation and Policy Analysis 21(4): 365-383. Morgan, Stephen L. (1998). "Adolescent Educational Expectations: Rationalized, Fantasized, or Both?" Rationality and Society 10(2):131-62. Morgan, Stephen L. (2005). On the Edge of Commitment: Educational Attainment and Race in the United States. Stanford: Stanford University Press. Need, Ariana and Uulkje de Jong (2001). "Educational Differentials in the Netherlands - Testing Rational Action Theory." Rationality and Society 13(1): 71-98. Oakes, Jeannie A. (1990)."Opportunity, Achievement and Choice: Women and Minority Students in Science and Mathematics." Review of Research in Education 16: 153–222. 27 Oakes, Jeannie A., Adam Gamoran and Reba N. Page (1992). "Curriculum Differentiation: Opportunities, Outcomes and Meanings." Pp. 570-608 in Handbook of Research on Curriculum edited by Philip W. Jackson. New York: Macmillan. Plank, David N. and Gary Sykes (2003). Choosing Choice: School Choice in International Perspective. New York: Teachers College Press. Schneider, Mark, Paul Teske and Melissa Marschall (2000). Choosing Schools: Consumer Choice and the Quality of American Schools. Princeton, NJ: Princeton University Press. Shavit, Yossi (1984). "Tracking and Ethnicity in Israeli Secondary Education." American Sociological Review 49(2): 210-220. Shavit, Yossi (1990). "Segregation, Tracking and the Educational Attainment of Minorities: Arabs and Oriental Jews in Israel." American Sociological Review 55(1): 115-126. Tel Aviv-Jaffa, Municipality. http://www.tel-aviv.gov.il/Hebrew/Education/ schools/Index.htm, accessed March 2006. 28 Table 1: Percent of Boys and Girls Who Choose to Study Subjects at Advanced Levels Subject Boys Girls Total Physics 33 12 22 Chemistry 22 17 19 Computers 28 12 20 Biology 19 20 19 Economics 30 25 28 Social Science 30 52 41 History 16 16 16 Literature 12 15 13 Communications 15 25 20 Table 2: Subjects by Perceived Utility and Perceived Risk Subject Physics Perceived Utility in Perceived University Risk of Admission Failure 4.35 22.2 Chemistry 4.01 15.2 Computers 4.14 12.4 Biology 3.83 9.7 Economics 3.91 9.7 Social Science 3.69 5.5 History 3.40 4.2 Literature 3.24 6.6 Communications 3.21 5.1 29 Table 3: Composition of Advanced Subjects Chosen Choice Composition Hard subjects only Soft subjects only Mix of hard and soft subjects Total a % 21 52 27 100 44 cases (6 percent) did not select a major and are excluded from the analysis. Table 4: Means (and their S.E.) of Key Variables by Choice of Subjects Variables Social Background Sex (Boys) All (1) Hard Subjects (2) Mixed (3) Soft Subjects (4) 0.50 (0.50) 0.69 (0.47) 0.58* (0.50) 0.38* (0.49) Parental education 4.68 (1.55) 5.25* (1.15) 4.16* (1.73) 4.63* (1.55) Economic resources 6.20 (2.19) 6.81* (1.86) 5.48* (2.18) 6.22* (2.26) 82.97 (10.76) 88.10* (8.41) 80.68 (11.74) 81.46* (10.56) 0.79 (0.41) 0.95* (0.22) 0.81* (0.40) 0.70* (0.46) 0.004 (0.98) -0.09* (1.06) 0.19* (0.95) -0.03 (0.95) 1.20 (0.27) 1.34* (0.31) 1.18 (0.25) 1.14* (0.24) 0.10 (0.25) 0.004 (0.15) 0.03* (0.19) 0.17* (0.29) 569 128 150 291 Academic Performance & Grouping Academic performance Grouping in math (advanced level) Secondary Factors Relative risk aversion Relative utility (hard sciences vis-à-vis soft subjects) Relative failure expectation (hard sciences vis-à-vis soft subjects) Number of Cases NOTE: Asterisks in column 2 indicate statistically significant (p<0.05) differences between the means shown in columns 2 and 3. Asterisks in column 3 indicate statistically significant (p<0.05) differences between the means shown in columns 3 and 4. Asterisks in column 4 indicate statistically significant (p<0.05) differences between the means shown in columns 2 and 4. 30 Table 5: OLS Regression Coefficients (and their S.E.) on Academic Performance and Secondary Factors. Independent Variables Performance and Grouping Secondary Factors Academic Performance Group^ Relative Risk Aversion Sex (Boys) -0.72 (0.78) 0.23* (0.06) 0.25* (0.07) 0.08* (0.02) -0.08* (0.02) Parental education 1.69* (0.29) 0.13* (0.02) -0.06* (0.03) 0.02* (0.01) 0.00 (0.01) Economic resources 0.99* (0.20) 0.03* (0.02) -0.05* (0.02) 0.01* (0.01) 0.01 (0.01) Academic performance 0.00 (0.00) 0.004* (0.00) -0.002* (0.00) Grouping in math 0.03 (0.10) 0.09* (0.03) -0.06* (0.03) (advanced level) Relative Utility Relative Failure Expectation Constant 69.29* (1.49) 1.31* (0.11) 0.25 (0.30) 0.58* (0.08) 0.35* (0.08) Adjusted R2 0.14 0.11 0.04 0.14 0.04 ^ Measured on a three-point scale corresponding to levels 3 (regular), 4, and 5 (advanced). * p<0.05 31 Table 6a: Primary and Secondary Effects (and their S.E.) on the Odds of Choosing Hard rather than Soft Subjects Independent Variables Social Background Sex (Boys) Hard Subjects vis-à-vis Soft Subjects II III IV V I 1.30* (0.21) 1.42* (0.22) 1.37* (0.22) Parental education 0.49* (0.13) 0.32* (0.14) Economic resources 0.17 (0.12) 1.06* (0.22) VI 1.40* (0.22) 1.19* (0.23) 0.17 (0.14) 0.16 (0.14) 0.13 (0.15) 0.06 (0.13) 0.06 (0.13) 0.05 (0.13) 0.01 (0.14) 0.86* (0.15) 0.82* (0.15) 0.82* (0.15) 0.70* (0.16) 1.69* (0.40) 1.67* (0.40) 1.42* (0.41) -0.09 (0.11) -0.11 (0.12) Academic Performance & Grouping Academic performance Grouping in math (advanced level) Secondary Factors Relative risk aversion -0.14 (0.11) Relative utility 0.72* (0.12) 0.52* (0.13) -0.86* (0.15) -0.88* (0.18) (hard sciences vis-à-vis soft subjects) Relative failure expectation (hard sciences vis-à-vis soft subjects) Constant Cox and Snell R2 -1.57* (0.17) -1.77* (0.18) -3.15* (0.40) -1.47* (0.18) -3.15* (0.40) -2.99* (0.42) 0.14 0.19 0.23 0.22 0.23 0.31 * p<0.05 32 Table 6b: Primary and Secondary Effects (and their S.E.) on the Odds of Choosing Hard Sciences, rather than Mixed Subjects Independent Variables Social Background Sex (Boys) I Hard Subjects vis-à-vis a Mix of Subjects II III IV V VI 0.52* (0.25) 0.62* (0.26) 0.60* (0.26) Parental education 0.64* (0.15) 0.49* (0.15) Economic resources 0.44* (0.14) 0.44 (0.25) 0.65* (0.26) 0.60* (0.26) 0.43* (0.15) 0.42* (0.15) 0.42* (0.16) 0.34* (0.15) 0.38* (0.15) 0.35* (0.15) 0.33* (0.15) 0.74* (0.16) 0.74* (0.16) 0.74* (0.17) 0.69* (0.17) 0.80 (0.44) 0.79 (0.44) 0.64 (0.45) -0.22 (0.13) -0.21 (0.13) Academic Performance & Grouping Academic performance Grouping in math (advanced level) Secondary Factors Relative risk aversion -0.33* (0.13) Relative utility 0.54* (0.13) 0.28* (0.14) -0.14 (0.17) -0.23 (0.19) (hard sciences vis-à-vis soft subjects) Relative failure expectation (hard sciences vis-à-vis soft subjects) Constant -0.32 (0.21) * p<0.05 33 -0.52* (0.22) -1.23* (0.45) -0.31 (0.21) -1.24* (0.45) -1.20* (0.46) Table 6c: Primary and Secondary Effects (and their S.E.) on the Odds of Choosing Soft rather than Mixed Subjects Independent Variables Social Background Sex (Boys) I Soft Subjects vis-à-vis a Mix of Subjects II III IV V VI -0.79* (0.21) -0.80* (0.21) -0.77* (0.21) Parental education 0.15 (0.11) 0.18 (0.11) Economic resources 0.27* (0.12) -0.62* (0.21) -0.74* (0.21) -0.59* (0.22) 0.26* (0.11) 0.25* (0.12) 0.29* (0.12) 0.28* (0.12) 0.32* (0.12) 0.30* (0.12) 0.32* (0.13) -0.12 (0.12) -0.08 (0.12) -0.07 (0.12) -0.02 (0.12) -0.89* (0.27) -0.88* (0.27) -0.79* (0.28) -0.13 (0.11) -0.10 (0.12) Academic Performance & Grouping Academic performance Grouping in math (advanced level) Secondary Factors Relative risk aversion -0.18 (0.11) Relative utility -0.18 (0.13) -0.24 (0.13) 0.72* (0.14) 0.65* (0.14) (hard sciences vis-à-vis soft subjects) Relative failure expectation (hard sciences vis-à-vis soft subjects) Constant 1.25* (0.15) * p<0.05 34 1.25* (0.15) 1.93* (0.26) 1.16* (0.16) 1.91* (0.26) 1.79* (0.28)
© Copyright 2026 Paperzz