Rational Educational Expectations?
A mixed methods study on background-related beliefs and preferences
M. Verhoeven 6047831
m-verhoeven@hotmail.com Research Master Social Sciences
First reader: Prof. Dr. H.G. van de Werfhorst Second reader: Dr. S.M. Steinmetz
June 30th, 2015 Amsterdam
Abstract
Dutch high school pupils from advantaged backgrounds more often enroll in higher levels of education than equally able pupils from less advantaged backgrounds. According to a sociological version of rational action theory, this may partly be explained by background-related beliefs and preferences. However, research on these presumably explanatory factors is sparse, which is why the present paper employs a mixed methods research design to gain more insight in these mechanisms. First, the effect of social background on pupils’ academic self-assessment is studied quantitatively, as well as the extent to which this mechanism explains the effect of social origin on pupils’ educational expectations. Additionally, a qualitative analysis of semi-structured interviews with high school pupils explores other forms of background-related beliefs and preferences. The results of the quantitative analysis demonstrate that, as expected, social background positively affects pupils’ academic assessment. Also, a positive effect is found of academic self-assessment on educational expectations. Yet, self-self-assessment only mediates the effect of social background on educational expectations to a minor extent. The interview analysis, on its turn, suggests that, instead, the following four possible background-related beliefs and preferences might help to explain persisting educational inequalities: the importance pupils attach to effort versus ability in education, pupils’ time discount rates, what pupils are looking for in a job, and parental preferences regarding pupils’ own education.
Keywords: Educational inequality, background-related beliefs, background-related preferences,
educational expectations, rational action theory, mixed methods
Introduction
Dutch society is increasingly open and meritocratic. More and more, people’s professional and educational careers are affected by their individual achievements rather than by their social background (see e.g. De Graaf & Luijkx 1992; Tolsma & Wolbers 2010: 22-6). Nevertheless, pupils from advantaged families are still more likely to develop and live up to higher educational expectations than pupils from less advantaged families, irrespective of their own intellectual capacities (Kloosterman et al. 2009). What remains unclear, though, is how these differences in expectations can be explained. Why are some more inclined than others to attain higher educational credentials? And to what extent is this a form of disguised educational inequality?
One of the theories that has been addressing these issues is a sociological version of rational action theory (see e.g. Breen & Goldthorpe 1997; Goldthorpe 1998). What makes this theoretical perspective sociological is that it moves beyond intentional actors who make fully-informed economic cost-benefit analyses by implementing three other core mechanisms: background-related beliefs, background-related preferences, and relative risk aversion. These mechanisms supposedly cause people to adopt different educational expectations (see especially Breen & Goldthorpe 1997; Goldthorpe 1998; Breen 1999). Whereas previous studies already provided a fair amount of evidence for the latter mechanism (see e.g. Need & De Jong 2000; Davies, Heinesen & Holm 2002; Van de Werfhorst & Hofstede 2007), much is still unknown about the former two. The present paper aims to fill this gap by providing new insights into both the possible effects and dimensions of these beliefs and preferences in order to further diminish educational inequalities.
Given the novelty of research on background-related beliefs and preferences, the present study applies a mixed methods research design as to obtain a comprehensive as possible picture of what these mechanisms entail. First, in addition to the limited amount of empirical research that is available on background-related beliefs (see e.g. Stocké 2007; Becker & Hecken 2009; Tolsma, Need & De Jong 2010), the effect of social background on an innovative measurement of academic self-assessment is quantitatively examined using the Trends in International Mathe-matics and Science Study (TIMSS) 2003 dataset. Also, it is studied to what extent differences in educational expectations held by people from distinct social backgrounds are mediated by pupils’ academic self-assessment. Second, semi-structured interviews with eleven high school pupils were conducted. In these interviews, other, still undiscovered background-related preferences and con-victions of information upon which pupils base their educational expectations are explored. The aim here is to improve future operationalizations of the concepts of background-related beliefs and preferences. Below, this paper continues by presenting a more extensive discussion of sociological rational action theory.
Theoretical background
Sociological rational action theory builds on Boudon’s distinction between primary and secondary effects (1974). Whereas primary effects comprise background-related inequalities in academic abilities, secondary effects refer to background-related inequalities in choices and expectations, given children’s academic ability. While recognizing the presence of primary effects, sociological rational action theory’s main concern is to explain the existence of secondary effects. It theorizes that pupils from different backgrounds, regardless of their demonstrated ability, develop different beliefs about their opportunities and constraints and, consequently, maintain different prefe-rences. This would be the reason why pupils establish different educational expectations, as will now be demonstrated on the basis of the Dutch context.
In the Netherlands, the education system comprises many transition and branching points. After pupils finished primary school in grade six, they are allocated to separate tracks in either the first or second year of high school. Generally, pupils’ age varies from twelve to thirteen by then. In total, there are four different preparatory vocational tracks belonging to the level ‘vmbo’, next to an intermediate general track (havo), and two preparatory tracks for university (vwo). What also characterizes the Dutch education system is that the financial costs for education are rather modest, regardless of the education program. Yet, there are significant diffe-rences in required time investments. While it is possible to attain an upper-secondary vocational degree at the age of seventeen, a graduate university degree requires at least five years of additional schooling. According to Breen and Goldthorpe (see especially Goldthorpe 1996; Breen & Goldthorpe 1997), leading figures in sociological rational action theory, the higher time costs of the relatively prestigious types of education are more likely to deter those from less advantaged backgrounds than those from more advantaged backgrounds. This is the case, they claim, because the risk of failing might perceived to be higher by the former group than by the latter, based on the prospects their social positions proved to offer in the past. As a consequence, people from different backgrounds are thought to develop different time discounting preferences: dependent on their social position, they might prefer safe short-term plans over more risky long-term options.
A key question is whether pupils from disadvantaged backgrounds indeed adopt, unconsciously and unintentionally, false beliefs and therefore preferences that reinforce their unequal position in the education system. Goldthorpe (1998) also argues, for instance, inspired by Popper (1972; 1976), that the very same academic achievements might be assessed more pessi-mistically by pupils from disadvantaged backgrounds than by pupils from more advantaged families, which causes them to develop less ambitious educational expectations. Also, academic self-assessment might have a more positive effect on the educational expectations of pupils from more advantaged backgrounds than on the expectations of pupils from less advantaged back-grounds. This will be examined in the quantitative part of the present paper.
Additionally, Breen contends that people modify their beliefs in light of the outcome of their actions (1999). Hence, beliefs might change over time, even when they prove to be strongly related to social background. A relevant example comprises various studies by Marsh who de-monstrates that pupils compare their own academic performances with those of other pupils in their class and base their academic self-concept, the perception of their own academic abilities, on this comparison (see e.g. Marsh & Parker 1984; Marsh 1987). Consequently, similarly able students have a lower academic self-concept and presumably develop lower educational expec-tations when they are surrounded by more able pupils, than when they are surrounded by less able ones: a so-called Big-Fish-Little-Pond effect occurs. Furthermore, in case pupils are allo-cated to a high school track that does not match their previous expectations, they can be expected to adjust their beliefs to their new situation. More in general, besides and in contrast to background effects, the idea is that the higher the track level pupils are in, the higher their academic self-assessment, and the more ambitious their educational expectations.
Also, Shavelson, Hubner and Stanton (1976) argue that individuals’ self-assessments are informed by their beliefs about others’ perceptions towards them. What Areepattamannil and Freeman (2008) derive from this, is that pupils from ethnic minorities might score lower on academic self-concept than native pupils. Moreover, this might, although to a lesser extent, also apply to women in comparison to men, especially in gender-stereotyped subjects like mathematics (see e.g. Meelisen & Luyten 2008). The aforementioned effects of class average achievements, high school track, ethnicity and gender on both academic self-assessment and educational expectations are controlled for in the quantitative analysis. However, to delimit the scope of the current paper, they are not extensively studied.
Finally, Breen argues that the probability of success in education depends on both individual effort and ability (1999). Departing from the assumption that pupils generally adopt their parents’ beliefs regarding the importance of effort and ability in education, he hypothesizes that the less advantaged might believe that one mainly needs ability rather than effort to be successful in education. This could cause these pupils to put little effort in their education from the start, which might result in low achievements that, in turn, confirm their beliefs and make these pupils give up on their educational career at an early stage. Moreover, when effort does prove to be important for a successful school career, this vicious circle would reproduce educational inequalities. The qualitative part of the current study will, among other things, focus on this possible background-related difference in the attached importance of ability and effort. In the next section, previous research findings on background-related beliefs and preferences are dis-cussed and hypotheses are established.
Previous empirical research
The fact that secondary effects are present in the transition to post-secondary education has been well-established by now (see for the Netherlands e.g. Tieben & Wolbers 2009; Kloosterman 2010), but, contrary to the mechanism of relative risk aversion, little is known about the mediating power of background-related beliefs and preferences. Thus far, research on background-related beliefs mainly focused on either the interaction between background and previous school career (Gambetta 1996; Breen & Jonsson 2000; Morgan 2005), or on subjective success probabilities, irrespective of previous educational careers (Stocké 2007; Becker & Hecken 2009; Tolsma, Need and De Jong 2010). Additionally, a study by Breen, Van de Werfhorst and Meier Jæger (2014) examined background-related time discounting preferences in connection to educational decision making after compulsory education. These studies will now be discussed more in detail.
Social background, school career and subjective success probabilities
Gambetta’s research (1996) focuses on educational decision-making after compulsory education in North-West Italy, and concludes that pupils base the decisions regarding their education on expected success probabilities. Moreover, Gambetta finds indicators for a positive relation be-tween an interaction of social class and school career on the one hand, and subjective success probabilities on the other. Failures in previous education appear to have a more severe negative effect on the educational choices made by working-class pupils than on those made by middle class pupils. However, this relation was not directly tested.
Breen and Jonsson’s study (2000) connects to Gambetta’s research by examining the Swedish case of alternative pathways in education, as an advancement of Mare’s sequential model (1980). In studying the effect of alternative pathways on later transition probabilities from one level or type of education to another, Breen and Jonsson find that pupils who were never situated in a pre-academic track are unlikely to eventually enroll in tertiary education. Furthermore, they find that the probability to enter tertiary education decreases the longer ago the enrollment in a pre-academic program was. Even more importantly, since a parallel could be drawn with the many possible trajectories in the Dutch education system, Breen and Jonsson explore social class effects on educational decision-making in alternative pathways. The results merely provide mixed evidence for such effects: social background appears to have a significant, positive effect on decision-making in some pathways, but not in others.
Next, Morgan (2005) attempts to understand why some adolescents in the USA enroll in post-secondary education whereas others do not. His results suggest that pupils’ educational attainment is affected by their background-related educational expectations, which are responsive to changes in incentives like education attitudes in their social surroundings.
What the three discussed studies have in common is, first, that they assume the presence of background-related subjective success probabilities without actually studying them. This is where the quantitative part of the present research steps in. Second, these studies have difficulties to explain why there (often) appears to be an interaction effect of social background and education career on educational decision-making. Consequently, to fill this gap, the qualitative part of the current paper will explore whether there are any other relevant background-related beliefs that might mediate this relation.
More recently, increasing attention is paid to the direct measurement of subjective success rates in education. For instance, Stocké (2007) analyzed whether there is a class effect on parental subjective success probabilities. These probabilities were measured by asking parents how likely they thought their child was to attain specific degrees. Also, it was explored whether parental subjective success probabilities could explain the effect of social class positions on educational careers. Surprisingly, the results of this German panel study demonstrate that differences in subjective success probabilities are class-related, yet suppressing factors that, moreover, are almost fully explained by primary effects rather than by secondary effects.
Additionally, Becker and Hecken (2009) explore the German case about high school pupils’ own educational decision to either enroll in vocational training or in higher education. What they find is that pupils from a higher class are more likely to enroll in higher education than pupils from a lower class. Also, Becker and Hecken asked pupils how they assess their performances at school and whether they believe that they are able to successfully complete their studies at university because of their previous schooling. What the results show is that, if primary effects are controlled for, both academic self-evaluation and subjectively expected likelihood of success at university mediate the effect of background on pupils’ educational decision-making.
Tolsma, Need and De Jong (2010) examine whether subjective estimates of success probabilities explain the effect of social background on students’ choices between different school tracks in Dutch higher education. They asked first year students to rate their subjective success probabilities in percentages for different courses at different levels of higher education. In line with the sociological rational action theory, Tolsma, Need and De Jong find that social background has a positive effect on these probabilities, irrespective of ability. However, no relation is found between subjective probabilities of success and level of higher education. Hence, in contrast to Becker and Hecken’s study (2009), the results of Tolsma, Need and De Jong do not indicate that subjective success probabilities mediate the effect of social background on track enrolment in higher education. As a consequence, while also taking into account Stocké’s fin-dings (2007), it remains difficult to unequivocally explain secondary effects in education.
Hypotheses
As is demonstrated above, previous research either studied subjective success probabilities indirectly or asked how pupils evaluate their academic performances. However, thus far subjective success probabilities have not been measured relative to pupils’ academic achieve-ments. Hence, it is difficult to examine whether similar school performances are actually assessed differently by pupils from different backgrounds, as is hypothesized by sociological rational action theory. Therefore, the quantitative part of the current study employs an innovative measurement of academic self-assessment, based on pupils’ academic self-concept, which will be put to the test by the following hypotheses:
1) Social background has a positive effect on high school pupils’ academic self-assessment.
2) Social background positively affects high school pupils’ educational expectations.
3) The positive effect of social background on educational expectations is mediated by pupils’ academic self-assessment.
4) Academic self-assessment has a more positive effect on the educational expectations of pupils from an advantaged social background than on the expectations of pupils from less advantaged backgrounds.
Time discounting preferences
Finally, moving from background-related beliefs to background-related preferences, Breen, Van de Werfhorst and Meier Jaeger (2014) find an effect of time discounting preferences on educational making, yet no mediating effect is found to explain differences in decision-making across students from different backgrounds. However, the used operationalization of time discounting preferences could be improved: Danish pupils moving to post-compulsory education were asked about their preferences for three hypothetical jobs, rather than education programs, with different economic returns and time horizons.
Design, data, and methods Research design
One of the major strengths and contributions of this paper is that the effect and shape of background-related beliefs and preferences are studied with a mixed methods research design. Such a design combines quantitative and qualitative research methods to bundle the strengths and minimalize the weaknesses of different methods in order to obtain the most comprehensive picture of the topic under study (Greene 2012; Johnson & Onwuegbuzie 2004). In the present study, an equal emphasis is put on both parts of the analysis that were performed synchronously, and the limited generalizability of qualitative research is partly complemented by the quantitative examination of academic self-assessment as one dimension of background-related beliefs. Simultaneously, the ecological validity pitfall which is often associated with quantitative research is accounted for and elucidated by the qualitative part. Together, the data allow for an elaborate analysis of both background-related beliefs and preferences.
It should be noted, though, that the interviews were performed in the spring of 2015, whereas the quantitative data stems from 2003. The reason why more recent TIMSS datasets were not used is that these datasets do not comprise information on Dutch eighth graders. However, this time gap is challenging, since, in the years between 2003 and 2015, Dutch student benefits changed from a gift to a loan. This transformation clearly has financial consequences for students, and might affect their educational expectations and ambitions. To cope with this data problem, all interviewees were asked whether they thought the recently introduced policy had changed their future education plans. None of them, though, felt this caused a revision of their aims and expectations. Nevertheless, combining quantitative and qualitative data from the same would have further improved the present study.
Data
In order to study the effect of social background on academic self-assessment and the way differences in educational expectations across people from distinct social backgrounds are mediated by this dimension of background-related beliefs, TIMSS 2003 data are used. This data-set is the result of an extensive international study on student achievement of both eighth and
fourth graders in mathematics and science, supplemented by student, teacher and school self-completion background questionnaires (Martin 2005). Assuming that fourth graders do not have well-developed educational expectations yet, and with the present paper’s scope being the Netherlands, solely TIMSS 2003 data on Dutch eighth graders are examined. In total, 2919 Dutch eighth graders from 128 different classes in 128 different schools participated in that study. The final sample for the current analysis comprised 1711 pupils.
Additionally, to explore still undiscovered background-related beliefs and preferences, eleven semi-structured in-depth interviews are performed with high school pupils from either the ninth, tenth, or eleventh grade of the intermediate general track (havo). Four of these interviewees are male and seven are female. The reason why respondents from this age and track are approached is that, first, younger pupils are, again, less likely to have already developed post-secondary educational expectations, whereas the selected pupils either have to decide upon a course profile or a post-secondary education program. Second, this group is most directly confronted with the largest amount of opportunities. Unlike pupils in the other tracks, pupils in the intermediate general track, at least theoretically, have to consider three possible options instead of two: intermediate vocational school, tertiary vocational college, and university. Pupils from the preparatory vocational tracks are not likely to consider a university degree as a realistic expectation, whereas pupils from the pre-university track are not thought to expect an intermediate vocational qualification as their highest degree. Because of this wider range of options, the background-related beliefs and preferences were most likely to come to the fore in discussing educational expectations with pupils in the intermediate general track.
The convenience sample of interviewees1 was established through the author’s social network, which allowed to recruit pupils in the intermediate general track on a Dutch high school. In total, 279 pupils of this high school were asked to fill out a brief survey on their parents’ education, their parents’ occupation, their own school achievements and their educa-tional expectations. Also, a question was included about whether the pupils wanted to participate in an interview, and, if so, they were requested to leave their contact information. Beforehand, parents were asked for their permission. Eventually, nineteen respondents left existing e-mail addresses or phone numbers, but, due to various reasons, only eight of these pupils could actually be interviewed. In addition though, three respondents from other high schools were interviewed as well. When both education level and occupation are taken into account, nine of these respon-dents can be considered to have a middle social background, whereas two - Ioana and Petr - are from a lower social background.
On average, the interviews lasted forty-five minutes. Ten out of eleven interviews were audiotaped and transcribed, and one respondent requested not to record the interview. Hence, this respondent’s answers were written down. The interviews themselves (mostly performed at school or in a quiet café), asked pupils about their educational career thus far, about their family situation, their parents’ expectations, their own subjective academic performance, their post-secondary plan and expectations, and, finally, about whether, why and when these plans and expectations changed in the past.
Quantitative operationalizations
The outcome variable of the first analysis is academic self-assessment. As a first step in developing this measure, an index was created to determine pupils’ academic self-concept. Due to data re-strictions, this variable is solely based on math and science subjects, and not on cultural and societal courses. Below, the eight Likert scale survey items that are used to construct this variable are listed. Each of these items has four answering categories varying from ‘totally agree’ to ‘totally disagree’: 1) I usually do well in math, 2) I usually do well in physics/chemistry, 3) I usually do well in biology, 4) I usually do well in earth sciences, 5) Mathematics is more difficult for me than for many of my classmates, 6) Physics/chemistry are more difficult for me than for many of my classmates, 7) Biology is more difficult for me than for many of my classmates, 8) Earth sciences are more difficult for me than for many of my classmates. The resulting index ranges from 1.5 to 4 and has a Cronbach’s alpha of .66, a reliability rate that is not perfect but is what realistically can be expected when measuring complex psychological constructs (Kline 1999). Subsequently, the index of academic self-concept is regressed on both pupils’ math and science performance (see ‘school achievement’ for these concepts’ operationalizations)2, and the variable academic self-assessment is derived from the residuals of this regression to measure the way pupils evaluate their own academic achievements. Thus, as a subjective element that is considered a core element of background-related belief differentiations, pupils’ self-concept is examined relative to their academic performance.
Educational expectations, the second dependent variable, represents respondents’ answers to
the question ‘How far in education do you expect to go?’. This variable includes the answering options 1) intermediate vocational school, 2) tertiary vocational college, and 3) university. The fourth category, ‘I do not know’ is recoded as missing.
In order to measure pupils’ social background, a variable is constructed for the highest attained education level of either parent. This variable is derived from two other variables that measure mother’s and father’s education separately, and the new variable includes the following values after recoding: 1) intermediate vocational school or less, 2) tertiary vocational college, and 3) university.
Additionally, because the key interest of this study is to examine secondary effects, primary effects are controlled for. As an indicator of academic achievement an extensive measurement of both pupils’ math and science skills is used. These achievements were measured by different items examining pupils’ skills in algebra, measurement, numbers, geometry and data (mathematics achievements), and in life science, chemistry, physics, earth science and environmental science (science achievements). In the current analysis, the standardized math and science scores – with a mean score of 50 and a standard deviation of 10, weighted for the booklet version filled out by the pupil (Martin 2005: 2-14) - are used as a measure of respondents’ school performance.
With respect to high school track, all the pre-vocational tracks are combined in a first category, while the intermediate general track forms the second category, a mixed intermediate general and pre-university track the third category, and the pre-university track the fourth
2 It is decided to use the residuals of the regression on both pupils’math and science performance rather than the residuals of regressions on either one of them or on their joint mean score, because this operationalization results into the best model fit (Adjusted R2 = .06). Hence, this results in the smallest residuals possible, which allows for the most rigid examination of the extent to which academic self-assessment is affected by social background.
category.
Next, two variables are created, of which one measures the class average math achievement, whereas the other measures the class average science achievement. In this way, a possible Big-Fish-Little-Pond effect can be controlled for.
Then, a dummy variable is constructed for gender, with male pupils being the reference category. Finally, when it comes to ethnicity pupils were coded as native when they themselves as well as both of their parents were born in the Netherlands. The other respondents were coded to be immigrants. The former group was taken to be the reference category.
Data strategy
In order to assess whether educational expectations vary across people from different social backgrounds, and to what extent this relation is mediated by academic self-assessment, multivariate regression analyses are performed. The effect of pupils’ social background on the way they evaluate their school performance is examined by OLS regression analyses. Next, ordinal logistic regression analyses of social background on educational expectations are conducted. Furthermore, because pupils who attend the same school might be confronted with similar circumstances that could affect their outcomes (for instance, the socioeconomic status of the people in their surroundings), the assumption of independent errors is violated, which causes exaggerated error terms and unjustly significant coefficients. Therefore, all analyses are performed with robust standard errors clustered by school to take the correlation between pupils, their classes and their schools into account.
To explore other dimensions of background-related beliefs, next to dimensions of background-related preferences, a qualitative thematic content analysis is conducted. This analysis started with a round of open coding, where apparently relevant parts of the data were labelled while looking for different types of beliefs and preferences upon which high school pupils base their educational expectations. After this first round of coding, the codes were examined more systematically, and, subsequently, different possible background-related preferences and beliefs were distinguished.
Quantitative analysis Descriptive Statistics
Table 1 presents the descriptive statistics. What strikes one first when studying this table is that pupils in the pre-vocational track generally score higher on academic self-assessment than could be expected based on their actual achievements, whereas pupils in all other tracks appear to underestimate their capabilities. This might be related to the Big-Fish-Little-Pond effect (Marsh & Parker 1984; Marsh 1987), but a further exploration of this lies outside the scope of the present study. What additionally can be derived from Table 1 is that most pupils in each track expect to eventually attain an educational degree that matches their current education level. Also, a fair amount of pupils has parents who completed a degree similar to the one the pupils appear to be heading for, at least based on their current track. This relation seems to be less strong, though, for pupils in the intermediate general/pre-university and pre-university tracks, which
Table 1: Descriptive statistics of pupils in each of the distinguished high school tracks, weighted for population size Variables Pre-vocational Mean (S.D.) Intermediate general Mean (S.D.) Intermediate general/ pre-university Mean (S.D.) Pre-university Mean (S.D.) Sample Mean (S.D.) Academic self-assessment Educational expectations
Intermediate vocational school
Tertiary vocational college
University
Social background (based on parental education)
Intermediate vocational school or less
Tertiary vocational college
University Math achievement Science achievement
Class average math achievement Class average science achievement Gender (Females) Ethnicity (Immigrants) .11 (.39) .54 (.50) .38 (.49) .08 (.28) .65 (.48) .25 (.43) .10 (.30) 43.86 (6.99) 44.81 (7.83) 43.86 (6.99) 44.81 (7.83) .45 (.50) .05 (.22) -.05 (.36) .19 (.39) .61 (.49) .19 (.40) .34 (.47) .50 (.50) .17 (.38) 52.85 (6.03) 52.84 (6.94) 52.85 (6.03) 52.84 (6.94) .45 (.50) .02 (.14) -.04 (.41) .13 (.34) .46 (.50) .40 (.49) .32 (.47) .38 (.49) .29 (.46) 56.48 (7.00) 55.63 (7.83) 56.48 (7.00) 55.63 (7.83) .52 (.50) .06 (.23) -.03 (.42) .02 (.14) .17 (.38) .81 (.39) .26 (.44) .36 (.48) .38 (.49) 59.60 (6.68) 58.69 (7.39) 59.60 (6.68) 58.69 (7.39) .55 (.50) .02 (.16) .02 (.40) .28 (.45) .37 (.48) .35 (.48) .44 (.50) .34 (.47) .22 (.42) 51.65 (9.62) 51.62 (9.66) 51.65 (9.62) 51.62 (9.66) .49 (.50) .04 (.20) N 669 182 407 453 1711 Source: TIMSS 2003
appears to be in line with the recent meritocratic trend that has been noticed in the Netherlands (see e.g. De Graaf & Luijkx 1992; Tolsma & Wolbers 2010: 22-6). Then, as can be expected, the average academic achievement of both pupils and their classes increases with every track. Furthermore, Table 1 shows that there are, on average, somewhat more men than women in the two lowest tracks, whereas the opposite applies to the two highest tracks. Lastly, no ethnicity pattern appears to be present across the different high school track. To continue, the upcoming paragraphs study the relationship between social background, academic self-assessment and educational expectations.
Social background and academic self-assessment
In Table 2 it is examined to what extent social background positively affects the way pupils evaluate their academic abilities based on their achievements. Model 1 demonstrates the direct effect of social background on pupils’ academic self-assessment, which is positive yet insig-nificant. In Model 2 high school track is included as well. Here, the effect of social background became significant. Pupils whose parents attained an intermediate vocational qualification or less have a significantly lower academic self-assessment than pupils whose parents completed a higher degree. No significant differences were found between children of parents educated at the three different tertiary levels. Hence, the effect of background is suppressed by high school track. It
Table 2: Coefficients of multivariate OLS regression analyses examining the effect of parental education on academic self-assessment
Variables Model 1
b (S.E.) Model 2 b (S.E.) Model 3 b (S.E.) Model 4 b (S.E.) Model 5 b (S.E.) Model 6 b (S.E.)
Social background (tertiary vocational college is ref. cat.)
Intermediate vocational school or less
University
High school track (intermediate general track is ref.cat.)
Pre-vocational track
Intermediate general/pre-university track
Pre-university track Class average math performance Class average science performance Gender (male is ref. cat.) Ethnicity (native is ref. cat.) Constant -.00 (.02) .03 (.03) .01 (.02) -.05* (.02) .05 (.03) .18*** (.04) .00 (.05) .00 (.04) -.04 (.03) -.05* (.02) .05 (.03) -.00 (.01) -.01 (.01) .66 (.13)*** .00 (.02) .03 (.03) -.17*** (.02) .10*** (.02) -.00 (.02) .02 (.03) .14** (.05) .01 (.02) -.05* (.02) .03 (.03) . .11* (.05) .04 (.05) .07 (.04) -.00 (.01) -.01 (.01) -.15*** (.02) .09 (.05) .39 (.20) Adjusted R2 N 1711 .00 1711 .04 1711 .04 1711 .04 1711 .00 1711 .09 Source: TIMSS 2003 * = p ≤ .05, ** = p ≤ .01, *** = p ≤ .001
should be noted, though, that the negative effect of a disadvantaged social background is rather small (-.05, p≤.05).
Next, Model 3 suggests that class average math and science achievements do not affect academic self-assessment. However, these variables also appear to suppress the effect of social background. The results in Model 4 and 5 demonstrate that both gender and an immigrant status significantly affect academic self-assessment, without really changing the direct effect of social background in Model 1. Women appear to develop less optimistic beliefs about their academic abilities than men (-.17, p≤.001). Surprisingly, immigrants have a significantly higher academic self-assessment than native pupils (.14, p≤.01). Finally, Model 6 presents the full model, where the negative effect of a relatively disadvantaged background is present again, and the significant positive effect of being an immigrant disappeared.
In sum, support is found for the first hypothesis regarding the positive effect of social background on pupils’ academic self-assessment. Hence, academic self-assessment appears to be a background-related belief, which is in line with earlier findings by Stocké (2007), Becker and Hecken (2009) and Tolsma, Need and De Jong (2010). It should be noted, though, that the explained variance in academic self-assessment is very limited, even in the full model (Adjusted R2=.09).
Academic self-assessment and educational expectations
Subsequently, it is explored whether educational expectations differ across people from different social backgrounds and to what extent these differences are mediated by pupils’ academic self-assessment. As can be derived from Model 1 in Table 3, a significant difference in educational expectations between people from different backgrounds is found. Pupils whose parents attained an intermediate vocational degree or lower are significantly less likely to expect a higher education level versus a lower one than pupils whose parents have a tertiary vocational degree (p≤.001). Pupils with parents who finished university have, on their turn, a 1.48 higher log odds (p≤.001) than pupils with parents who attained a tertiary vocational degree to expect a higher education level.
Next, Model 2 controls for primary effects, high school track, class average achievements, gender and ethnicity. This allows to examine the mediating effect of academic self-assessment in Model 3. What this model demonstrates is that academic self-assessment indeed has a positive effect on educational expectations. With every unit of increase in pupils’ academic self-assessment, the log odds of expecting a higher education degree rather than a lower one increases by .84 (p≤.001). However, academic self-assessment only explains background-related differences in educational expectations to a minor extent: the positive effect of having parents who finished university versus having parents who have a tertiary vocational degree, as well as the negative effect of having parents who completed intermediate vocational school or less versus having higher educated parents hardly change when Model 2 and 3 are compared. Moreover, the explained variance in educational expectations only increase by .01 when academic self-assessment is added to the model.
Finally, from Model 4 it can be derived that interactions of social background and academic self-assessment do not significantly affect the expectations that high school pupils developed. Also, the inclusion of these interaction terms does not improve the model fit. It
should be noted, though, that adding these interaction terms to the model causes the positive effect of academic self-assessment to increase from .84 (p≤.001) to 1.01 (p≤.001).
Table 3: Coefficients of ordinal logistic regression analyses examining the effect of parental education on educational expectations
Variables Model 1
b (S.E) Model 2 b (S.E.) Model 3 b (S.E) Model 4 b (S.E.)
Social background (tertiary vocational college is ref. cat.)
Intermediate vocational school or less
University Math achievement Science achievement
High school track (intermediate general track is ref. cat.)
Pre-vocational track
Intermediate general track/pre university track
Pre-university track Class average math achievement Class average science achievement Gender (male is ref. cat.) Ethnicity (native is ref. cat.)
Academic self-assessment
Social background intermediate vocational degree x academic self-assessment Social background university degree x academic self-assessment -1.15*** (.12) 1.48*** (.17) -.71*** (.13) 1.34*** (.19) .03** (.01) .02* (.01) -.52* (.26) .37 (.23) 2.20*** (.29) .05 (.04) -.03 (.03) -.18 (.12) 1.78*** (.38) -.69*** (.14) 1.31*** (.19) .02 (.01) .02* (.01) -.61* (.26) .36 (.22) 2.18*** (.29) .07 (.04) -.02 (.03) -.09 (.12) 1.66*** (.38) .84***(.15) -.68*** (.14) 1.31*** (.19) .02 (.01) .02* (.01) -.62* (.26) .36 (.23) 2.18*** (.29) .07 (.04) -.02 (.03) -.09 (.12) 1.65*** (.38) 1.01***(.23) -.30 (.30) -.25 (.45) Pseudo R2 N 1711 .11 1711 .30 1711 .31 1711 .31 Source: TIMSS 2003 * = p ≤ .05, ** = p ≤ .01, *** = p ≤ .001
In sum, the results support hypothesis 2 and 3. Social background appears to positively affect high school pupils’ educational expectations, and seemingly this effect can partly be explained by academic self-assessment. It should be noted, though, that the beliefs that pupils develop about their academic abilities only appears to play a modest mediating role. Furthermore, the results do not indicate a more positive effect of academic self-assessment on the educational expectations
of high school pupils from a more advantaged social background than on the expectations of pupils from less advantaged backgrounds, which rejects hypothesis 4.
Qualitative analysis
Now that the quantitative analysis showed that academic self-assessment hardly mediates the differences in educational expectations across pupils from different backgrounds, the interview analysis presents some possible background-related beliefs and preferences that might be more helpful in explaining secondary effects. Below, in Table 4, the respondents are introduced.
Table 4: An overview of the interview respondents’ characteristics3
Repondent Age Sex Grade Father’s education Mother’s education Father’s occupation Mother’s occupation
Diego 15 Male 10 University Intermed. voc.
school Civil servant Nurse
Samantha 15 Female 10 Tertiary voc.
college Tertiary voc. college furniture business Owner of a Financial staff member Ioana 16 Female 11 Intermed. voc.
school Pre-vocational track Team manager Supermarket staff member Louise 17 Female 11 University Tertiary voc.
college Unknown Volunteer
Johan 16 Male 10 Tertiary voc.
college Pre-vocational track medical business Owner of a Nurse
Petr 16 Male 10 None Intermed. voc.
school Warehouse staff member Sales staff member
Jess 15 Female 10 University University ICT staff member Self-employed
teacher Sophia 14 Female 9 Tertiary voc.
college University Engineer Cleaner
Anna 18 Female 10 Tertiary voc.
college Intermed. voc. school Managing director Self-employed consultant Fleur 17 Female 10 University Tertiary voc.
college Entrepreneur Teacher
David 15 Male 10 Tertiary voc.
college Tertiary voc. college Self-employed copywriter Actress
Ability versus effort
As is hypothesized by Breen (1999), beliefs about effort and ability are an often recurring theme in the interview analysis. Moreover, these beliefs appear to be informed by pupils’ previous school career, which is in line with earlier studies by Gambetta (1996), Breen and Jonsson (2000), and Morgan (2005). For instance, Samantha, Ioana and Louise descended from the pre-university track to the intermediate general track, which convinced all three of them, despite their different
3 This table presents self-reported information. Sometimes, parents’ occupations was reported in rather
abstract terms like ‘civil servant’, because the respondents were not fully informed about their parents’ specific function.
backgrounds, that it mainly takes ability, rather than effort, to complete a high degree. As an example, Louise stated:
“I wanted it so badly [continuing at the pre-university level], and I worked as hard as I could, but my test results just weren’t good enough.”
Due to their descending experience, it seems that none of them feels they are as able as they think they have to be for the pre-university track and this caused them to adjust their educational expectations.
In contrast, Johan and Petr’s experiences led them to believe that one mainly needs effort to succeed in education. Johan says he made it to the tenth grade at the inter-mediate general level without showing any effort and hence he concludes when asked what one needs to go to university:
“Perseverance, a lot of it. And a lot of discipline. […] Yes, I think these are the two main things you need at uni.”
Furthermore, he does not seem bothered by the fact that neither of his parents are academically educated.
Petr, who comes from a relatively low background, first finished the pre-vocational track in high school, without ever having to study for an exam. Currently, he is enrolled in the tenth grade of the intermediate general track and he feels he needs to “step it up” in terms of effort, which implies that ability only is no longer sufficient to succeed in education. Yet he is convinced that predominantly effort is required to reach one’s goals. For instance, he replies to the question how large he considers his chances to attain a university degree:
“I think that as long as I am driven, and I really really want it, if I find something that I truly enjoy, I think I can make it.”
Nevertheless, Petr does not fully downplay ability’s role:
“I think that if you have the brains and enough perseverance, you should be fine [in attaining a university degree]. But if you’re just not smart enough, some people just aren’t , we aren’t all equally bright, then it might be very hard.”
It is difficult to tell, though, whether this can be explained by Petr’s background.
The citations above are merely some examples, for all interviewees made claims regarding the role of effort and ability in education. What can be derived from these examples, though, is that there are differences across pupils in the importance they attach to ability and effort. Also, these beliefs appear to be at least partly shaped by pupils’ previous school experiences. However, what remains unclear is to what extent these different beliefs are background-related. In the present analysis no convincing indicators were found for such a relation. Nevertheless, since abi-lity and effort form such a prevalent theme in the analysis, it is deemed important to further study whether pupils’ social origin affects the importance they attach to effort and ability, and to what extent this explains educational inequalities.
Time discounting preferences
Although time discounting preferences are less widespread in the interviews than beliefs regarding the importance of ability and effort, it is something that appears to affect high school pupils’ educational expectations. To give an example, Petr responded, when asked whether he could picture himself at university:
“Well, I’d rather work and be done with school as quickly as possible”.
This might be related to his relatively disadvantaged background. Additionally, Jess answers to the question why she does not aim to complete a university degree:
“Well, I think… I mean, tertiary vocational college is already 4 years. I think I’ve had enough of it by then.”
Moreover, she has this time preference despite the fact that, first, both her parents went to university, and, second, she assesses her abilities high enough to attain a university degree.
What is remarkable as well, is that both Petr and Jess’s time discounting preferences are joined by their aim to complete at least a tertiary vocational degree. Such a degree is not the fastest to attain, yet takes one to two years less than an academic qualification. Hence, they appear to be concerned with not staying in education any longer than is necessary to attain a level that they deem acceptable.
Opposite time preferences are found as well. They come to the fore most clearly in Sophia’s story. Sophia, mother went to university, aspires a neurosurgeon degree, which will take her at least eight more years in education. Yet, she does not appear to mind this prospect:
“If it [the study] is fun and I really like it, I think it won’t be too hard time-wise. These six years [after two additional years to get a pre-university degree] will pass by before I know it.”
Like Petr’s time discounting preference, Sophia’s time discounting preference appears to be background-related too.
In sum, there appear to be differences in time discounting preferences. Whereas some are mainly occupied with finishing their education as fast as they can, others are willing to invest a lot of time in education. Furthermore, unlike Jess’s example, Petr and Sophia’s cases indicate that these preferences are background-related. Yet, more research is required to examine whether and to what extent time discounting preferences mediate the effect of social background on educational expectations.
Money or fun? Job and study preferences
A third theme in the interviews was what the pupils were looking for in their future jobs and education programs. Remarkably, Anna, neither of whose parents attained an academic qualify-cation, hopes to one day get a master’s degree, because:
“It really adds something and it lifts you to a higher level […] It really means something to belong to this small group of people who made it that far in education.”
Hence, when it comes to education, Anna appears to be in search for status. Yet, this is not something she is aiming for in her job:
“With a career I don’t feel like that. I don’t aim for the highest possible as to… it is not my goal, let’s put it like that. It’s just not my goal.”
When she is asked whether she aims for something else in a job, she replies: “I want to have a fun job, one which is not too easy, but just fun.”
Most of the other respondents, irrespective of their social background, share Anna’s job view. Fleur, who could be considered to come from a higher middle background, for instance, mentions:
“I notice more and more that writing is really for me. I also know that you’re actually supposed to find a job which is fun and well-paid, but that, then again, is not for me […] What I find important in life is happiness and for that you don’t need a good job.”
Additionally, David, who comes from a middle background, seems to agree with Anna’s view on education:
“Yes, I find it important to be extremely good at something, and to have everything, all the information there is… That I know everything. […] I want to aim for the highest level possible.”
However, both David and Diego disagree with the others’ job views, since they find a well-paid job at least as important as a job that they enjoy. When they are invited to talk about what kind of job they want, they both immediately mention “a job with an above average income”. Apart from the fact that these different jobs and study preferences appear to be gender-related, it would be interesting to further explore whether they explain differences in educational expectations across pupils from different social backgrounds.
Parental preferences
Finally, there might be important differences in parental preferences when it comes to pupils’ education level. Whereas Anna’s and Ioana’s parents do not express any preferences, others, like Sophia’s mother (who went to university), do:
“So, my mother has this very persistent idea of ‘Sophia go to university’ and, and ‘go get a good job’ […] My mom really supports the idea of me becoming a neurosurgeon. She always was already like ‘Sophia, become a doctor, or an engineer like your father’, or ehm, what was it again? Oh, yes, a lawyer.”
It seems plausible that pupils whose parents have high expectations and make them explicit, feel the urge to meet their parents’ expectations. What if Sophia’s mother would not express her preferences? What if they were less high? Would Sophia still aim for such a prestigious university degree?
Also, despite Louise’s transition from the pre-university track to the intermediate general track, Louise’s mother keeps expressing her support for a second attempt to attain a (pre-)
university degree, while she herself completed a tertiary vocational degree. Although Louise, in the major part of her interview, appeared to be quite convinced that tertiary vocational school would suit her best as well, she admitted that, sometimes, she still struggles with the question whether or not to try to meet her mother’s expectations.
An important thing to note here is that parental preferences concerning children’s educational attainment do not necessarily correspond to parents’ own education levels. Johan’s father, for instance, was never enrolled in a (pre-)academic program, but he explicitly pushes Johan to aim for at least a pre-university degree. Therefore, the mechanism of parental preferences appears to differ from the relative risk aversion mechanism, although it is possible that parental preferences take potential credential inflations into account, which would be in line with findings from Van de Werfhorst’s research (2009). Another possibility is that social background affects parental preferences in the sense that the less advantaged might be less likely to have high expectations regarding their children’s education than people in a more advantaged position, regardless of the fear for status demotion. This would have to be studied further.
In total, four possible mechanisms that might explain secondary effects are distilled from the performed interviews. It is now up to future research to further examine which of these background-related beliefs and preferences are indeed able to explain differences in educational expectations across pupils from different social backgrounds in order to gain new insights in how persistent educational inequalities can be tackled.
Conclusion and discussion
Although education has become increasingly accessible in the Netherlands, educational inequalities do persist. As is demonstrated in the current paper once again, equally able pupils from advantaged social backgrounds are more likely to develop higher educational expectations to live up to than pupils from disadvantaged backgrounds. In order to explain this phenomenon, the present paper departed from a sociological rational action theory and explored the possible effects and dimensions of both background-related beliefs and preferences.
In doing so, this mixed methods study adds to previous research in two ways. First, the quantitative part examined the innovative mediating mechanism of academic self-assessment, which measured academic self-concept relative to high school pupils’ actual school performance. Subsequently, it was examined to what extent this mechanism explains differences in educational expectations across pupils from different backgrounds. Second, the qualitative analysis enabled to gain momentum in identifying and operationalizing other possible dimensions of background-related beliefs and preferences, which allows future research to rely less on theory and more on inductive data.
The results of the quantitative analysis indicate that high school pupils’ academic self-assessment is background-related, and does, moreover, positively affect pupils’ post-secondary educational expectations. Nevertheless, the mediating effect that was found is very limited. Hence, other background-related beliefs and preferences are likely to be present, and these were explored by the qualitative part of the study. Besides beliefs regarding the importance of effort and ability to succeed in education, and the ways these beliefs are informed by pupils’ previous school career, three potentially background-related preferences were distilled from the data: time discounting preferences, preferences concerning study and job features, and parental preferences regarding their children’s educational attainments.
Inevitably, the present study has some weaknesses. As was mentioned before, there is a considerable time gap between the quantitative data from 2003 and the interview data from 2015. It would be wise for future research to reduce this time gap, for instance, by collecting new quantitative data that, moreover, incorporate information about the newly established possible forms of background-related beliefs and preferences. Also, the measurements of academic achievement and academic self-assessment should be improved in future research. Because the TIMSS 2003 dataset does not include information on cultural and societal subjects, these variables were measured with information on mathematics and science self-concepts only. However, including data on additional courses is likely to improve the explanatory power of both constructs.
Thirdly, apart from background-related beliefs and preferences, deficiencies in more objective information – for instance regarding the actual level of academic courses - may also cause differences in educational expectations. Furthermore, it would be interesting to explore whether other indicators of social background, like social class, would change the results. However, the TIMSS 2003 data do not allow to study these latter two matters. Finally, the effect of academic self-assessment, as well as the distinguished beliefs and preferences, may vary across branching or decision points in the education system, but this too could not be examined. Therefore, scholars are invited to take both the results and the aforementioned weaknesses of this study to heart, in order to make further progressions in explaining secondary effects to tackle remaining educational inequalities.
Despite the limitations discussed above, two policy implications can be drawn from the current study in case further research finds evidence for the newly established mechanisms in the qualitative part of the present study. First, to counter any inequality-reinforcing preferences that are caused by people’s background (and hence not by cultural mechanisms), it is important to provide all high school pupils with as much qualitative and quantitative information as possible on the pay offs of different jobs, studies, and education levels. This will enable people from relatively disadvantaged backgrounds to make educational choices based on information that will be more similar to the information of people from a more advantaged background.
The second policy implication concerns not only the interaction between pupils’ previous school career and background-related beliefs about the importance of effort and ability, but also parental preferences regarding their children’s education. When these mechanisms turn out to mediate the effect of social background on educational expectations, it would, in case of doubt about pupils’ academic ability, be best to aim high in order to prevent educational decision-making from being biased by background-related factors that reinforce educational inequalities. The trajectories that pupils take in the education system should be informed by demonstrated rather than expected academic abilities. This will allow everyone to fully develop themselves, and then we might achieve a truly meritocratic society.
Acknowledgements
This paper would not have existed without the extensive help of Herman van de Werfhorst. I would like to thank him, as well as Stephanie Steinmetz for their generous comments. Furthermore, I would like to thank Paul, Joke, Michelle, Eva, Sterre, Minke and Estrelle for their mental support. Needless to say, I feel very grateful towards the high school that allowed me to recruit the interview respondents. Finally, I owe special thanks to the interviewees, who were so kind to spend their time on this research project that is so dear to me.
References
Areepattamannil, S., and Freeman, J.G. 2008 ‘Academic Achievement, Academic Self-Concept, and Academic Motivation of Immigrant Adolescents in the Greater Toronto Area Secondary Schools’, Journal of Advanced Academics 19(4): 700-43.
Becker, R., and Hecken, A.E. 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(1): 25-45.
Boudon, R. 1974 Education, Opportunity, and Social Inequality: Changing Prospects in Western Society, New York: Wiley.
Breen, R. 1999 ‘Beliefs, Rational Choice and Bayesian Learning’, Rationality and Society 11(4): 463-79.
Breen, R., and Goldthorpe, J.H. 1997 ‘Explaining Educational Differentials: Towards a Formal Rational Action Theory’, Rationality & Society 9(3): 275-305.
Breen, R., and Jonsson, J.O. 2000 ‘Analyzing Educational Careers: A Multinomial Transition Model’, American Sociological Review 65(5): 754-72.
Breen, R., Van de Werfhorst, H.G., and Meier Jæger, M. 2014 ‘Deciding under Doubt: A Theory of Risk Aversion, Time Discounting Preferences, and Educational Decision-making’, European
Sociological Review 31(2): 144-60.
Davies, R., Heinesen, E., and Holm, A. 2002 ‘The relative risk aversion hypothesis of educational choice’, Journal of Population Economics 15: 683-713.
Gambetta, D. 1996 Were They Pushed or Did They Jump? Individual Decision Mechanisms in Education, Boulder, CO: Westview Press, Inc.
Goldthorpe, J.H. 1996 ‘Class Analysis and the Reorientation of Class Theory: The Case of Persisting Differentials in Educational Attainment’, The British Journal of Sociology 47(3): 481-505. Goldthorpe, J.H. 1998 ‘Rational action theory for sociology’, The British Journal of Sociology 49(2): 167-92.
Graaf, P.M., de, and Luijkx, R. 1991 ‘Van ascription’ naar ‘achievement’? Trends in status-verwerving in Nederland tussen 1930 en 1980’, Mens en Maatschappij 67(4): 412-33.
Greene, J.C. 2012 ‘Engaging critical issues in social inquiry by mixing methods’, American
Behavioral Scientist 56: 755-73.
Johnson, R.B., and Onwuegbuzie, A.J. 2004 ‘Mixed Methods Research: A Research Paradigm Whose Time Has Come’, Educational Researcher 3(7): 14-26.
Kline, P. 1999 The handbook of psychological testing, London: Routledge.
Kloosterman R. 2010 Social background and children’s educational careers: The primary and secondary effects
of social background over transitions and over time in the Netherlands, Nijmegen: Radboud University
Nijmegen.
Kloosterman, R., Ruiter, S., Graaf, P.M., de, and Kraaykamp, G. 2009 ‘Parental education, children’s performance and the transition to higher secondary education: trends in primary and secondary effects over five Dutch school cohorts (1965–99), The British Journal of Sociology 60(2): 377-98.
Mare, R. D. 1980 ‘Social Background and School Continuation Decisions’, Journal of the American
Statistical Association 75: 295-305.
Marsh, H.W. 1987 ‘The Big-Fish-Little-Pond Effect on Academic Self-Concept’ Journal of
Educational Psychology 79(3): 280-95.
Marsh, H.W., and Parker, J.W. 1984 ‘Determinants of Student Self-Concept: Is it Better to Be a Relatively Large Fish in a Small Pond Even if You Don’t Learn to Swim as Well?’, Journal of
Personality and Social Psychology 47(1): 213-31.
Martin, M.O. 2005 TIMSS 2003 User Guide for the International Database, Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College.
Meelissen, M., and Luyten, H. 2008 ‘The Dutch Gender Gap in Mathematics: Small for Achievement, Substantial for Beliefs and Attitudes’, Studies in Educational Evaluation 34(2): 82-93. Morgan, S.L. 2005 On the Edge of Commitment. Educational Attainment and Race in the United States, Stanford, CA: Stanford University Press.
Need, A., and Jong, U., de 2000 ‘Educational Differentials in The Netherlands. Testing Rational Action Theory’, Rationality and Society 13(1): 71-98.
Popper, K.R. 1972 Objective Knowledge, Oxford: Clarendon Press.
Popper, K.R. 1976 ‘The Logic of the Social Sciences’ in: T.W. Adorno (ed.) The Positivist Dispute in
German Sociology, London: Heinemann.
Shavelson, R.J., Hubner, J.J., and Stanton, G.C. 1976 ‘Self-concept: Validation of construct interpretations’, Review of Educational Research 46: 407-41.
Stocké, V. 2007 ‘Explaining Educational Decision and Effects of Families’ Social Class Position: An Empirical Test of the Breen-Goldthorpe Model of Educational Attainment’, European
Tieben, N., and Wolbers, H.J. 2009 ‘Transitions to post-secondary and tertiary education in the Netherlands: a trend analysis of unconditional and conditional socio-economic background effects’, Higher Education, 60: 85-100.
Tolsma, J., Need, A. and Jong, U., de 2010 ‘Explaining Participation Differentials in Dutch Higher Education: The Impact of Subjective Success Probabilities on Level Choice and Field Choice’, European Sociological Review 26(2): 235-52.
Tolsma, J. and Wolbers, H.J. 2010 Naar een open samenleving? Recente ontwikkelingen in sociale stijging en
daling in Nederland, Den Haag: Raad voor Maatschappelijke Ontwikkeling.
Werfhorst, H.G., van de 2009 ‘Credential inflation and educational strategies: A comparison of the United States and the Netherlands’, Research in Social Stratification and Mobility 27: 269-284. Werfhorst, H.G., van de, and Hofstede, S. 2007 ‘Cultural capital or relative risk aversion? Two mechanisms for educational inequality compared’, The British Journal of Sociology 58(3): 391-415.
