Whatever parents want?
Research on the role of parents’ ambitions in ethnic differences in children’s educational ambitions
Suzanne de Leeuw (1019883) Master thesis, final version Supervisor: Matthijs Kalmijn Second reader: Stephanie Steinmetz
2 Abstract
Earlier research shows that migrants have on average higher educational ambitions than their native peers. There are two actors influencing children’s ambitions present in the literature: schools and parents. On the one hand, the Bayesian learning theory suggests that positive school feedback, such as high grades and high track placement, results in higher ambitions. On the other hand, the theory of significant others and the social capital theory hypothesize that the ambitions of parents are the most important determinant of the educational ambitions of children. The role of both actors in the
aforementioned gap in educational ambitions between migrants and natives is investigated in this research. Additionally, we investigate the assumption, underlying the social capital theory, that parents successfully communicate their ambitions to their children. For this research the Dutch sample of the CILS4EU dataset has been used in which students (N=4363) and parents (N=3248) were asked about their educational ambitions. The rich data contain multiple ways to measure educational ambitions which makes it possible to deal with a number of methodological issues. The theories have been tested with a multinomial logistic regression. Firstly, we find a suppressor effect of school feedback: controlled for feedback, the gap in educational ambitions between migrants and natives is larger. Secondly, we show that parental ambitions explain fifty percent of the gap in educational ambitions between migrant and native students.
Introduction
Not every person aspires the highest goal possible. Instead some people have higher educational ambitions than others. Those differences in ambitions are relevant because research indicates that high ambitions positively affect educational performance (Sewell et al., 1969; Sewell & Hauser, 1980; Morgan, 1998; Feliciano & Rumbaut, 2005). One of the findings of the literature on educational ambitions is that first and second generation migrants in Western countries have on average higher educational ambitions than their native peers (See Heath et al (2008) for an overview). The literature names two actors influencing the educational ambitions of children: schools (e.g. Morgan, 2005) and parents (e.g. Coleman, 1988). This research investigates the role of both actors in the gap in
educational ambitions between children with and without an immigrant background.
Firstly, the Bayesian learning theory states that positive feedback from schools, such as high grades and high track placement, positively affect the educational ambitions of children. However, these factors do not seem to explain the high ambitions of migrants relatively to their native peers. Instead, earlier research shows that children with an immigrant background perform on average worse in education while their educational ambitions are on average higher (e.g. Jonsson & Rudolphi, 2011; Van de Werfhorst & Tubergen, 2007). Based on these findings a suppressor effect could be expected: the gap in educational ambitions between migrants and natives might be larger when we control for educational performance. A second part of the literature, represented by the theory of significant others and social capital theory, emphasizes the importance of parents’ ambitions for children’s educational ambitions. These theories state that high parental ambitions result in higher educational ambitions among their children. There are suggestions in the literature, such as formulated in the
3 family mobilization theory, that migrant parents have on average higher ambitions for their children than native parents. If the influence of parents is as large as the theory of significant others and the social capital theory suggest, this difference in ambitions between migrant and native parents could explain the gap in educational ambitions between children with and without an immigrant
background. However, this is never empirically tested before. Therefore, this research investigates in how far parents’ ambitions explain the gap in educational ambitions between students with and without an immigrant background.
The outcomes of this research are important for society and academia. First, it is very important that teachers know where the ambitions of their students originate from. This will make it possible to help students with extraordinary high or low educational ambitions during their educational career. Secondly, this research fills a gap in the academic literature on educational ambitions. Earlier research on the relation between educational ambitions of parents’ and children is limited. Most of the existing research is conducted decades ago and does not pay attention to ethnic differences (e.g. Haller & Portes, 1973; Rumbaut, 1995). Those who do take ethnic differences into account state that some ethnic groups are more ambitious than others (e.g. Gutpa, 1997 ) or that the influence of parental ambitions is not equally strong in all ethnic groups (e.g. Kerckhoff & Campbell, 1977; Hout & Morgan, 1975). However, these studies do not test the theory that the gap in ambitions between migrants and natives is the result of the different ambitions of their parents. Therefore, this research fills a gap in the literature.
In addition to the main question, this study contributes with a test of an assumption underlying the theory of significant others and the social capital theory. These theories assume that parents communicate their ambitions to their children. However, it is uncertain whether this assumption is correct. Therefore, this research does not only investigate self-reported ambitions of parents and children but also children’s estimation of their parents’ ambitions. This makes it possible to investigate whether the perceptions of children are accurate and whether children adapt their ambitions based on the wishes of their parents.
We use the CILS4EU dataset for this research. This dataset is unique because parents and children did both answer questions about their educational ambitions. Additionally, children did not only answer questions about their own ambitions but also about their parents’ ambitions. The multiple ways to measure educational ambitions make important methodological improvements in comparison to previous research possible.
4 Theory and previous research
We distinguish two important actors influencing children’s educational ambitions: schools and parents. The most important theories about the influence of those actors on educational ambitions are discussed in this section. Per actor we start with a description of their influence on educational ambitions in general. Subsequently, we discuss the special position of migrants in those theories. Schools: Bayesian learning theory
The Bayesian learning theory argues that the educational ambition of children are mainly the result of feedback by the school (e.g. Morgan, 2005). Schools constantly assess the skills of their students and communicate their findings to children and parents. The Bayesian learning theory states that children adapt their ambitions during the educational career based on the information they receive from the school about their abilities. Examples of feedback from the school are grades or track placement. Empirical tests of this theory confirm that grades (Stinebecker & Stinebecker, 2012; Zafar, 2011) and tracking (Buchmann & Dalton, 2002) are important determinants of children’s ambitions. Based on the Bayesian learning theory it could be expected that positive feedback from schools results in higher educational ambitions whereas negative feedback lowers the educational ambitions of children (hypothesis 1a).
There are reasons to assume that migrants and natives react differently to feedback. Eearlier research shows that children with an immigrant background perform on average worse in education while their educational ambitions are on average higher (e.g. Jonsson & Rudolphi, 2011; Van de Werfhorst & Tubergen, 2007). This is not in line with the Bayesian learning theory which
hypothesizes that children with lower performance have also lower ambitions. In other words, school feedback does probably not explain the high ambitions relatively to their native peers. Instead, a suppressor effect could be expected. We expect that the gap in ambitions between migrants and natives is larger when the educational performance is taken into account (hypothesis 1b). Parents: Significant other and social capital
The role of parents in the ambitions of their children was mostly investigated by sociologists in the 1960’s, 1970’s and 1980’s. In the heydays, theories such as the theory of significant others and the social capital theory came into existence.
The theory of significant others is part of the first wave of literature on the link between parents’ and students’ educational ambitions. The theory was formulated as a reaction on the status attainment model of Blau and Duncan (1967). Blau and Duncan concluded that parents transmit their
socioeconomic position relatively more often via education while the direct transmission of parents’ socioeconomic positions onto children declined after industrialization. Sewell et al. (1969:83) argued
5 that, Blau and Duncan failed to indicate why and how parents were able to transmit their
socioeconomic position via education. Therefore, Sewell et al. (1969) introduced the Wisconsin status attainment model in which they emphasized the role of ambitions. In this alternative model, significant others such as parents, teachers and friends, influence the educational attainment of children via educational ambitions (ibid, 1969). Significant others influence the ambitions and aspirations of the child and the child’s ambitions in their turn positively affect their educational attainment.
A very similar argument has been made by the social capital theory (Coleman, 1988). This theory assumes that parents are only able to transmit their socioeconomic position via education as a result of the presence of within family social capital. Within family social capital is generally operationalized with variables measuring the involvement and ambitions of parents (Coleman, 1988). The social capital theory signalizes, just as the theory of significant others, a positive indirect effect of parents’ ambitions and involvement on the educational attainment of their child via the educational
ambitions of children. The main difference between both theories is that the theory of significant others focuses on the role of parents’ ambitions, whereas the social capital theory puts more emphasis on parental involvement.
Based on both theories, a large amount of empirical research was published. The results support the idea that significant others such as parents (e.g. Hout & Morgan, 1975; Sewell & Shah, 1968) and friends (e.g. Haller & Butterworth, 1975; Woelfel & Haller, 1971) define students’ ambitions. Moreover, parental involvement by offering help, taking active interest in the children (e.g. Lareau, 1989; Rumberger et al., 1990), discussion of school-related matters (e.g. Ho & Willms, 1996; Rumberger et al., 1990) and an emphasis on good study habits (e.g. Clark, 1983) have found to be beneficial for children’s educational ambitions and attainment. Based on both theories and the empirical evidence of previous research, we expect that higher parental ambitions result in higher educational ambitions among children whereas lower parental ambitions result in lower educational ambitions among children (hypothesis 2a).
The family mobilization theory suggests that the ambitions of migrated parents are on average higher than the ambitions of native parents. Zeroulou (1988) discovered in his study on Algerian working class-families the importance of the existence of a social mobility project in the family. Migrant families are on average more ambitious and have a more proactive attitude towards
education because they tend to believe that their decision to move to another country contributes to upward mobility (ibid, 1988). The existence of such a mobility project among migrants is explained by the idea that migrants are not a representative sample of the population in their home country
6 (e.g. Borjas, 1990; Feliciano, 2005). The decision to migrate asks a considerable amount of ambition, perseverance and resources and only people with sufficient motivation will migrate eventually. (e.g. Borjas, 1990; Feliciano, 2005). As a result, migrants tend to have higher ambitions for their children and have a more proactive attitude towards education. If parents influence the ambitions of their children as social capital suggests, the high ambitions of migrated parents in comparison to native parents can explain the gap in educational ambitions between students with and without an immigrant background. In other words, the positive effect of immigrant background on the educational ambitions of children is mediated by the educational ambitions of parents (hypothesis 2b).
An implicit assumption inherent to all the theories on the effect of parental ambitions on the ambitions of children, is that parents communicate their ambitions successfully to their children. Hypothesis 2a and 2b are both built upon this assumption. It is a crucial step in the theoretical reasoning of this research. However, there is little empirical evidence present in the literature to support this premise. Therefore this issue receives extra attention in the analysis. We label the assumption that parents communicate their ambitions to their children the bridge hypothesis because the assumption is as a bridge that leads to hypothesis 2a and 2b.
Methodology Dataset
This research uses the CILS4EU dataset. This dataset contains a sample of 14-year old children in Sweden, Great-Britain, Germany and the Netherlands who were questioned in 2010. The CILS4EU dataset is unique for two of reasons. Firstly, children and parents received both their own
questionnaire and answered questions on various topics such as language use, networks, and educational ambitions. Secondly, as a result of the oversampling method, the data are suitable for a minority-majority comparison.
We chose to use only the Dutch sample for this research. The Dutch case is interesting because the education system in the Netherlands is highly stratified. Students are selected into different tracks at the age of 12 and this track placement is an important determinant for the rest of their educational career. This makes the Dutch case ideal to investigate the differences in educational ambitions between migrants and natives in a setting where feedback by the school plays a major role. Sampling and non-response
The data has been collected with a three-stage and school-based sampling strategy. First, schools with students in the target age (14 years) were selected. Within each selected school, children and their parents within two randomly selected classes were approached for the survey. To reach
7 parents, students were asked to provide one of their parents with an self-completion questionnaire. In the Netherlands 100 schools, 222 classes, 4363 students and 3 248 parents participated in the survey. For the oversampling strategy, schools in the sampling-frame were assigned to four mutually exclusive strata based on the proportion of students with an immigrant background. Schools in the strata with a larger proportion of students are oversampled.
To limit the non-response, a replacement strategy was used. When a school in the random sample did not want to participate, a school with similar characteristics on explicit stratification criteria was approached to take part in the research instead. In other words, a school in a similar stratum and geographical region was used to replace a non-participating school. Unfortunately, the non-response was lower among students and parents with an immigrant background. Nevertheless, due to the effective oversampling method, 34% of the students and 26% of the parents have an immigration background.
For the purpose of this research some respondents have been removed from this initial sample because they did not answer crucial questions. Altogether, after the removal of the missing values, the sample consists of 3 146 students and 2 387 parents. For the regression analysis the sample of 2 387 students whose parents answered all relevant question has been used. The large sample of 3 146 students is only used for a sensitivity analysis and the descriptive statistics in figure 2. We use the small sample to avoid missing values in the parent level variables. However, the disadvantage is that the non-response in this sample is larger. An analysis of the excluded respondents in both samples shows that students with an immigrant background and students in the lower school types are more often removed from the sample (table 1). However, it looks like the non-response does not affect the average ambition level. The average ambitions of migrants and natives are in the samples of 3 336 and 2 530 students virtually the same and the ambitions in the original sample are only 0,1 point higher. Nevertheless probability weights have been used to limit the effects of the skewed non-response. Appendix A presents a detailed description of the used weights.
[TABLE 1 HERE]
Methods
For this research multinomial logistic regression analysis has been used with the educational
ambitions of children as dependent variable. For the regression analysis the sample of 2 378 children whose parents participated in the survey and answered the questions on ambitions has been used.
8 However, the non-response, especially among migrants, is very high in this sample. Therefore, a sensitivity analysis will be conducted to test whether the results are affected by the decline in sample size.
To compare estimates over models, the KHB-method has been used because the change in estimates between models in logistic regressions cannot be automatically attributed to the inclusion of new variables. The problem is that the estimates are influenced by omitted variables while those omitted variables vary between models (Mood, 2010). Kristian, Holm and Breen (2012) developed a
technique to make a comparison of estimates between models possible (KHB-method). This new method holds scale and error distributions constant over models. Moreover, a test has been developed to establish whether the change between models is significant.
Variables
The Dutch education system is visualized in figure 1. The Dutch names of the school types are placed in parentheses. After primary school, at the age of 12, students are split into three tracks: a pre university track (VWO), pre higher vocational track (HBO) and a pre lower vocational track (VMBO). We distinguish two types in the pre lower vocational tracks. First there is the so called VMBO basis/kader. This is the lowest type. Secondly there is VMBO g/t which is a bit more advanced. We distinguish between those two pre lower vocational school types because there are meaningful differences between the students with regards to non-response and ambitions. When students graduate, they can automatically continue in the matching track in post-secondary or tertiary
education. For example, students who graduate from the pre university track are allowed to continue in university. In addition students are allowed to accumulate diplomas (“stapelen”). This means that students who finished for instance a pre lower vocational track are allowed to continue in a pre higher vocational track.
The students in the sample are 14 years old and therefore already placed in one of the secondary school tracks. The current school type of students is included as a variable into the analysis. In this variable the lowest version of pre lower vocational education (VMBO basis/kader) functions as the reference category.
[FIGURE 1 HERE]
The dependent variable in this research measures the educational ambitions of children. To measure educational ambitions children were asked: “What level of education do you wish to get?” The
9 answer categories are “VMBO basis/kader”, “VMBO g/t”, “HAVO”, “VWO”, “MBO”, “HBO” and “University”. All variables on ambitions of students and their parents have the same answer categories.
Additionally a number of independent variables are important. There are two independent variables to measure school feedback. The first is the aforementioned variable measuring the current school type. Secondly, grades are included to measure feedback by the school. Grades are measured with students’ self-report grades on mathematics, Dutch and English. The highest of those three grades is used to measure educational performance.
There are also two variables to measure parents’ ambitions. First, parents were asked “What level of education do you wish your child to get?” The second option is to ask children about their parents’ ambitions. To measure children’s perception of parental aspirations, students were asked: “What level of education do your parents want you to get?” The answer categories of those items are very similar to the answer categories of the educational ambitions of children. The only difference is the inclusion of the answer category “don’t know.” This category is included to limit the non-response and because it is theoretically relevant. Theory suggests that parents communicate their ambitions to their children. However, this theory does not hold if parents don’t know what their ambitions are or if children don’t know what their parents want. For this reason it is relevant to include this category in the analysis.
Another important independent variable measures whether a student is considered to be a migrant or not. Until now the concept migrant is used loosely. In this research we focus on first and
secondary migrants from non-west countries1. Western migrants are not included in the sample. If the student or at least one of the parents was born abroad, the student is considered to be a
migrant. Based on this definition around 24% of the sample has an immigrant background. 5% of the respondents is born abroad (1,5 generation), 14% has two foreign parents (2nd generation) and 5% has one foreign and one native parent (2,5 generation). To avoid over-simplified images of the differences in educational ambitions between migrants and natives, the descriptive analysis will also display differences in ambitions between smaller ethnic categories. Appendix B contains more detailed information on those ethnic groups
Finally, gender (boy = 1) and socioeconomic background are included in the models as control variables. Parents’ occupational status reported by the child is used as an indicator of socioeconomic
1 Based on definition of Statistics Netherlands non-west migrants are migrants coming from Africa, Latin-America, Asia (excl. Japan and Indonesia) and Turkey.
10 background. The variable is recoded into the ISEI-scale. This measure is preferred over occupational status reported by the parents since the non-response among parents is high. Engzell and Jonsson (2015) have shown that selective non-response of parents is a serious problem which could cause biases in the data. However, children’s report on the occupation of their parents is often correct and the non-response is lower (ibid, 2015).
Analysis
The analysis consists of three parts. First, the data is introduced and the differences in ambitions between migrants and natives are described into detail. The second part focuses on the bridge hypothesis and investigates the assumption that parents communicate their ambitions successfully to their children. Finally, the regression analysis is presented in which the four hypotheses of this research are tested.
Migrants and natives
The starting point of this research is the finding that migrants have on average higher educational ambitions than natives (e.g. Van de Werfhorst & Tubergen, 2007). However, more detailed knowledge about this gap in ambitions is scarce. This section introduces the dataset and describes the differences in ambitions between migrants and natives into detail. For this purpose figure 2 has been added. The figure displays a number of things that should be discussed. Firstly, the figure confirms, in line with earlier research, that migrants have on average higher educational ambitions than their native peers. Migrant students and parents in all school tracks have higher educational ambitions than natives. Naturally, it could be expected that there are also differences in ambitions between migrant groups. However, a study of more detailed ethnic groups reveals that all migrant groups, except for the Antilleans, have higher educational ambitions than natives. The Antilles are officially a part of the Kingdom of the Netherlands. Many Antilleans speak Dutch and are more than other migrants familiar with the Dutch education system. It could be that this explains why their ambitions are also similar to those of native students
Secondly, figure 2 tells us something about the ambitions of students per track. Remarkably, the largest gap in ambitions between migrants and natives is situated in the lower tracks. 10,4% of the native students in pre lower-vocational tracks (VMBO) wish to get a university degree while 39,6% of the migrants in those tracks have the same aspiration. This results a gap of 29,2%. In contrast, the gap in ambitions between migrants and natives in the highest track is only 12,7%. 83,3% of the natives who are currently in the pre university track (VWO) wishes to get a university degree while 96,0% of the students with an immigration background in the pre university (VWO) track wishes the same. Additionally, there is a clear difference in ambitions between migrants in the lower and higher
11 version of the pre lower vocational track (VMBO basis/kader vs VMBO g/t). Migrants in the lower version have more often a university ambition while migrant children in the higher version of the pre lower vocational track wish more often for a higher vocational degree. It could be that students in the higher version genuinely believe that they have a chance to continue in the higher vocational track while the children in the lower version see no realistic chance to climb on the educational ladder. As a result, their wish for a university degree could be more based on ultimate dreams, bragging or social desirability than realistic expectations.
Thirdly, the figure contains information about the ambitions of parents. Parents with an immigrant background are on average more ambitious than native parents. When the ambitions of parents and children are compared, it becomes clear that native parents are less ambitious than their own children regardless the school type of the child. This is not in all tracks the case for migrant parents. Migrant parents with children in the higher version of the pre lower vocational track (VMBO g/t) have on average higher aspirations than their children. A final notable outcome of the figure is that native parents have a strong preference for higher vocational education (HBO). In contrast, children and migrant parents prefer university education.
[FIGURE 3 HERE]
Bridge hypothesis
The theory of significant others and the social capital theory implicitly assume that parents communicate their ambitions to their children. This assumption has been named the bridge hypothesis in the theory section. The advantages of the CILS4EU dataset are that parents and children are questioned independently about their ambitions and that children did also answer questions about their perceptions of parental ambitions. As a result it is possible to look deeper into this assumption. For this purpose table 3 compares the self-reported ambitions of parents and children while table 4 compares the self-reported ambitions of parents to children’s perceptions of parental ambitions.
First, it should be noted that 25% of the students say that they do not know what their parents want. This means that parents do not always communicate their ambitions successfully to their children. However, after the exclusion of children and parents who used the answer category “don’t know”, the correlation between what parents want and how children perceive their parents’ wishes, is 0,40. A very similar pattern is visible when we compare students’ ambitions with the ambitions of their
12 parents. The correlation between ambitions of parents and children is 0,43. These findings show that there is a moderate to strong connection between the ambitions of parents and students. However, we should be careful with these conclusions because there are reasons to believe that the reality is more nuanced as a result of clustering in the answer categories.
When the aspirations of parents and children are investigated further, it becomes clear that the answer categories are clustered. If parents report that they want their child to get a secondary school degree (pre lower vocational, pre higher vocational or pre university), the answer categories of their children are clustered in the same category or in the matching school type in post-secondary or tertiary education (respectively lower vocational, higher vocational or University). For example when parents want their children to get a pre higher vocational degree (HAVO), 35% of the children has the same wish but 33% wants to get a higher vocational degree (HBO). It is likely that parents and
children who aspire a HAVO or HBO degree want the same but formulate their ambitions differently. We label those differences short and long term ambitions. Students have short term ambitions when they report a secondary school level as highest ambition while students who report a post-secondary or tertiary school level as ambition have long term ambitions.
[TABLE 3 AND 4 HERE]
Since the data is clearly clustered, the matching school types will be combined in the regression analyses. This results in three answer categories: VMBO or MBO (lower vocational), HAVO or HBO (higher vocational) and VWO or University (university). Furthermore, extra dummy variables will be included to measure the difference between short (=0) and long term (=1) ambitions of parents and students. When the answer categories are clustered like this, the correlation between the ambitions of children and parents rises to 0,61. This means that the agreement between parents and children about ambitions is larger than it seemed to be at first sight.
In summary, the results to test the bridge hypothesis are mixed. On the one hand, the results show that 25% of the children does not know what their parents want. This indicates that parents do no communicate their ambitions effectively. On the other hand, if children think that they know what their parents want, the correlation between ambitions of parents and children is relatively high. This means that the assumption underlying the social capital theory and theory of significant others is valuable. However, we should be aware of the fact that not all parents successfully communicate their ambitions and that the theory does not match reality completely.
13 Regression analysis
The four hypotheses of this research have been tested in a multinomial logistic regression (table 5). The first model is a basic model which includes only immigrant background and control variables. The results confirm that migrants in the Netherlands have significantly higher educational ambitions than natives. For example, migrants are 3,52 times more likely to aspire a university degree
(VWO/University) and 2,2 time more likely to aspire a higher vocational degree relatively to a lower vocational degree (VMBO/MBO) than natives.
The second model tests the influence of school feedback. We formulated the expectation that positive feedback leads to higher ambitions among students in hypothesis 2a. The results confirm the hypothesis. Two types of feedback are added to the model: grades and the current school type. The grades have a small but significant effect: higher grades result in higher educational ambitions. For each grade point higher the chance that students aspire a higher vocational degree (HAVO/HBO) relatively to a lower vocational degree(VMBO/MBO) increases with the factor 1,3 and the chance to aspire a university degree increases with the factor 1,5. The effect of the currents track is also in the expected direction and significant. Children are the most likely to aspire the school type that matches with their current track. For example, children who are currently in the pre higher vocational track (HAVO) aspire mostly to get a higher vocational track (HAVO/HBO). The only odd result in this pattern is the insignificant effect of the pre university track. This is probably an error due to the limited sample. The sensitivity analysis in table 6 shows that, when the larger sample is used, the effect of pre university education becomes significant. Children in the pre university track most likely aspire a university degree. This is in line with the idea of the Bayesian learning theory that children adapt their ambitions based on the feedback from the school. Track placement is a very strong signal for students and as a result they adapt their ambitions accordingly.
With the addition of feedback variables in model 2, the effect of migration background increases. This is in line with the expectation, formulated in hypothesis 2b, that there is a suppressor effect. However, unlike in linear models, the change in estimates between models in logistic regressions cannot be automatically attributed to the inclusion of new variables. The problem is that the estimates are influenced by omitted variables which vary between models (Mood, 2010). Kristian, Holm and Breen (2012) developed a technique to make a comparison of estimates between models possible (KHB-method). This new method holds scale and error distributions constant over models. The results of this method confirm the increase in the estimate of immigration background between
2 Relative risk ratio
14 model 1 and 2. The effect of immigration background becomes 22% larger after the addition of feedback variables and this change is significant. These results confirm hypothesis 2b.
The last three models test the effect of parental ambitions on the educational ambitions of children. There are two ways two measure the educational ambitions: their self-reported ambitions and the perceptions of children of their ambitions. In model 3 the self-reported ambitions of parents are added. The results show a clear connection between the ambitions of parents and children. Higher parental ambitions result in higher educational ambitions among children. However, it looks like children are more ambitious than their parents. When parents aspire a university degree for their children, the chance that children aspire the same relatively to a lower vocational degree
(VMBO/MBO) increases with factor 20. However, children whose parents aspire a higher vocational degree are not equally likely to have the same dream as their parents. Instead they often have the same dream or a higher ambition (university degree) than their parents. Moreover, children rarely have a lower ambition than their parents. This indicates that children are generally more ambitious than their parents. Finally, if parents used the answer category “don’t know” the aspirations of their children are not significantly different from the ambitions of children whose parents aspire a lower vocational degree. In other words, the ambitions children with parents who don’t know what level of education they aspire for their child, have relatively low ambitions.
The fourth model replaces self-reported ambitions of parents with the perceptions of children. This measure has very similar effects. The main difference is that, with the use of perceptions, children are less likely to have higher ambitions than their parents. Additionally, when children report that they don’t know what their parents want, their ambitions are also relatively low. Altogether, those results confirm hypothesis 2a because they demonstrate the positive influence of high parental ambitions on the educational ambitions of children. The fifth model includes both indicators of parental ambitions simultaneously. Interestingly, both measures have still a significant effect when they are together in one model. Apparently, both measures do not measure precisely the same but have an additive and independent effect.
Hypothesis 2b states that parents ambitions explain the gap in educational ambitions between children with and without an immigrant background. To test this hypothesis the KHB-method is used again. The results of this approach show that, controlled for all variables in model 2, between 30% and 41% of the differences in ambitions between migrants is explained when either one of the measures of parental ambitions (self-reported or reported by the children) is added to the model. When both models are at the same time added to the model 50% of the differences in ambitions between migrants and natives are explained. Moreover, the effect of immigration background is not
15 significant anymore. These results are in line with hypothesis 2b and proof that parental ambitions are a very important explanation for the differences in educational ambitions between migrant and native students.
[TABLE 5 HERE]
For this regression analysis only the respondents whose parents participated in the survey were included. This results in a relatively small sample (N=2530) with a relatively high non-response, particularly among migrants. For some models in the regression analysis only student-level variables are used. In these models, it is possible to include students whose parents did not participated in the survey. This results in a larger sample with less non-response (N=3146). A comparison between the results of the small (N=2387) and large (N=3146) sample helps to discover whether the non-response in the small sample affected the results. For model 1, 2 and 4 a comparison between the reduced and large sample is possible. The sensitivity analysis is presented in table 6. The most important finding is that the overall conclusion is the same for both samples. This means that the exclusion of students whose parents did not answer the questions has not affected the overall conclusions. However, there are some small differences. Firstly, we mentioned already that the positive effect of a pre-university track becomes significant in the larger sample. Secondly, the effect of immigration background is bigger in the large sample. However, the overall trend is the same: there is a suppressor effect visible after the inclusion of feedback effects and the effect of immigration background reduces in model 4 after the inclusion of children’s perceptions. Finally, parental ambitions explain an equal amount of the gap in educational ambitions between students with and without an immigrant background when either of the samples is used.
[TABLE 6 HERE]
Discussion and conclusion
This research confirms that non-west migrants in the Netherlands have significantly higher
educational ambitions than their native peers. The influence of two actors on this gap in educational ambitions between migrants and natives has been investigated: schools and parents. Results indicate that positive feedback from schools, such as high grades and high track placement, positively affect the educational ambitions of children. However, it does not explain the high ambitions of migrants
16 relatively to their native peers. Instead, a suppressor effect was found: the ambitions of migrants relatively to natives are higher when educational performance is taken into account. This means that migrants lower their ambitions less based on negative feedback from the school than natives who receive a similar type of feedback. Additionally, the outcomes of this research report a positive effect of parental ambitions on children’s educational ambitions. High parental ambitions result in higher educational ambitions among children. More importantly, the findings of this research show that parental ambitions explain 50% of the gap in educational ambitions between migrants and natives. The theoretical literature has regularly named the high ambitions of migrant parents as an
explanation for the high educational ambitions of children with an immigrant background. The family mobilization theory predicts the high ambitions of parents and the theory of significant others and social capital theory emphasize the importance of these high ambitions among parents for the educational ambitions of children. Despite the fact that this line of reasoning is common in the theoretical literature, this research is the first to test and confirm it empirically.
In addition we investigated the underlying assumption of the theory of significant others and the social capital theory that parents communicate their ambitions to their children. The data show that the assumption is certainly valid. The high correlation between children’s perceptions of their parents ambitions and their parents self-reported ambitions points out that children are often aware of the wishes of their parents. During this investigation we discovered that some children and parents have long term ambitions while others have short term ambitions. Those who have long term ambitions formulate their ambitions in terms of post-secondary and tertiary education whereas those with short term ambitions formulate their ambitions in terms of secondary school types. For example, it is likely that a person who wishes for a pre university degree wants the same as a person who wishes for a university degree. The only difference is that one looks further ahead than the other. We discovered that the correlation between the ambitions of parents and children is much higher when we control for the differences in short and long term ambitions. The discovery of the importance of long and short term ambitions is an important methodological contribution of this research to the literature. Despite the high correlation, the assumption underlying the theory does not completely matches reality. A quarter of the students says that they are not aware of their parents’ wishes. This does not mean that assumption of which the theory of this research is build, is worthless. However, it is important to be aware of those mismatches between theory and reality. It would be valuable if future research finds out who the parents and children are who are unaware of each other’s ambitions and what factors determine whether parents are able to communicate their ambitions successfully to their children.
17 Due to the rich dataset of the CILS4EU this research was able to compare different measures of parental ambitions. We distinguish two ways to measure parental ambitions: self-reported parental ambitions and the estimates of parental ambitions as reported by their child. Results show that both measures have an additive and independent effect on the educational ambitions of children. We don’t know any other research that made this comparison between measures before and it is an important finding that both measures supplement each other. This is a methodological improvement and it makes our measure of parental ambitions stronger than other measures of parental ambitions used in previous research. There are a number of possible explanations for this finding. First, it could be that both measures gauge another aspect of parental support. For example, children’s perception of their parents’ ambitions influence their own ambitions because they want to live up to their parents’ expectations. However, those perceptions do not represent necessarily the actual ambitions of parents when parents do not communicate their ambitions explicitly. Even if children are unaware of their parents’ ambitions, they can be still influenced by them in an indirect manner. For example, because parents with high ambitions help more often with homework or are more involved in the educational career of the child. This line of reasoning could explain the independent effect of both measures of parental ambitions. Another possible explanation is a measurement error. There is always a certain error when measures are built upon one single survey question. Therefore, it is not surprising that two items measuring parental ambitions, which are based on questions to different persons, form a better measure than just one item.
This study has a number of limitations which should be discussed. First, the non-response is an issue. The non-response is clearly higher among students and parents with an immigrant background. However, we did not find any proof that this non-response has biased the results. The ambitions of students are in all samples virtually the same and the sensitivity analysis did not detect any major differences in outcomes. Nevertheless, the non-response should be taken seriously. Forster (under review) studied the non-response of the CILS4EU into detail and shows which characteristics of respondents, besides immigration background, contribute to non-response. This knowledge should be used to improve the sampling methods of future datasets. We noticed that not only the initial non-response is higher among migrants but unanswered questions within the survey are also more common among migrants. Secondly, we paid little attention to differences between migrant groups. This did not fall in the scope of this research but should get more attention in future research. It is likely that there are important differences between migrants based on ethnicity. In addition it could be that there are differences between migrants based on their reasons to migrate. For example, the ambitions of economic migrants may differ from the ambitions of refugees (e.g. Ogbu, 1992). Finally, the findings of this research have raised a new question. We discovered that the ambitions of
18 children are partly the result of their parents ambitions. This means that the high ambitions of migrant students are partly the result of the high ambitions of migrant parents. These results leads us to the next question: Why are the expectations of migrant parents this high? The family mobilization theory suggests that the high ambitions of migrant parents are the result of a family mobilization project in the family. However, an empirical research on the reasons for the high ambitions of parents is still missing in the literature.
In summary, this research has contributed to the research field in many ways. Most importantly, it shows how parental ambitions are an important explanation for the high ambitions of migrant students relatively to their native peers. This explanation was often mentioned in theoretical
discussions but rarely empirically investigated. Therefore, this research is a relevant contribution the academic literature. Moreover, the outcomes of the research contain important information for teachers and other people involved in the guidance of migrants students during their educational career. If they know where the ambitions of migrant students originate from, they will be better able to advice and help their students. This research is just a start and we summarized only a number of challenges that are left. Further research is needed to get a better understanding of the relatively high ambitions of children and parents with an immigration background.
19 Bibliography
Blau, P., & Duncan, O. (1967). The American occupational structure. New York: Wiley. Borjas, G. (1990). Friends or strangers. New York: Basic Books.
Buchmann, C., & Dalton, B. (2002). Interpersonal Influences and Educational Aspirations in 12 Countries: The Importance of Institutional Context. Sociology Of Education, 75(2), 99.
http://dx.doi.org/10.2307/3090287
Coleman, J. (1988). Social Capital in the Creation of Human Capital. American Journal Of Sociology, 94, S95-S120. http://dx.doi.org/10.1086/228943
Engzell, P. & Jonsson, J. (2015). Estimating Social and Ethnic Inequality in School Surveys: Biases from Child Misreporting and Parent Nonresponse. European Sociological Review, 31(3), 312-325.
http://dx.doi.org/10.1093/esr/jcv005
Erikson, R., & Jonsson, J. (1996). Can education be equalized?. Boulder, Colo.: Westview Press. Feliciano, C. & Rumbaut, R. (2005). Gendered paths: Educational and occupational expectations and
outcomes among adult children of immigrants. Ethnic And Racial Studies, 28(6), 1087-1118.
http://dx.doi.org/10.1080/01419870500224406
Forster, A. (unpublished) Non-response of parents in the CILS4EU survey: An examination of the influence of student, parents and school characteristics on sample selectivity.
Goyette, K., & Xie, Y. (1999). Educational Expectations of Asian American Youths: Determinants and Ethnic Differences. Sociology Of Education, 72(1), 22. http://dx.doi.org/10.2307/2673184
Gupta, Y. (1977). The Educational and Vocational Aspirations of Asian Immigrant and English School-Leavers--A Comparative Study. The British Journal Of Sociology, 28(2), 185.
http://dx.doi.org/10.2307/590210
Haller, A., & Butterworth, C. (1960). Peer Influences on Levels of Occupational and Educational Aspiration. Social Forces, 38(4), 289-295. http://dx.doi.org/10.2307/2573036
Haller, A., & Portes, A. (1973). Status Attainment Processes. Sociology Of Education, 46(1), 51. http://dx.doi.org/10.2307/2112205
Heath, A., Rothon, C., & Kilpi, E. (2008). The Second Generation in Western Europe: Education, Unemployment, and Occupational Attainment. Annu. Rev. Sociol., 34(1), 211-235.
20
http://dx.doi.org/10.1146/annurev.soc.34.040507.134728
Hout, M., & Morgan, W. (1975). Race and Sex Variations in the Causes of the Expected Attainments of High School Seniors. American Journal Of Sociology, 81(2), 364-394.
http://dx.doi.org/10.1086/226078
Jonsson, J., & Rudolphi, F. (2011). Weak Performance--Strong Determination: School Achievement and Educational Choice among Children of Immigrants in Sweden. European Sociological Review, 27(4), 487-508. http://dx.doi.org/10.1093/esr/jcq021
Karlson, K., Holm, A., & Breen, R. (2012). Comparing Regression Coefficients Between Same-sample Nested Models Using Logit and Probit: A New Method. Sociological Methodology, 42(1), 286-313. http://dx.doi.org/10.1177/0081175012444861
Kerckhoff, A., & Campbell, R. (1977). Black-White Differences in the Educational Attainment Process. Sociology Of Education, 50(1), 15. http://dx.doi.org/10.2307/2112641
Lareau, A. (1989). Home advantage. London: Falmer Press.
Mood, C. (2010). Logistic Regression: Why We Cannot Do What We Think We Can Do, and What We Can Do About It. European Sociological Review, 26(1), 67-82.
http://dx.doi.org/10.1093/esr/jcp006
Morgan, S. (1998). Adolescent Educational Expectations: Rationalized, Fantasized, or Both? Rationality And Society, 10(2), 131-162. http://dx.doi.org/10.1177/104346398010002001
Morgan, S. (2005). On the edge of commitment. Stanford, Calif.: Stanford University Press.
Ogbu, J. (1992). Understanding Cultural Diversity and Learning. Educational Researcher, 21(8), 5-14.
http://dx.doi.org/10.3102/0013189x021008005
Rumbaut, R. (1995). 'The New Californians: Comparative Research Findings on the Educational Progress of Immigrant Children. In R. Rumbaut & W. Cornelius, California's Immigrant Children (1st ed., pp. 17-69). Center of Mexican studies.
Rumberger, R., Ghatak, R., Poulos, G., Ritter, P., & Dornbusch, S. (1990). Family Influences on Dropout Behavior in One California High School. Sociology Of Education, 63(4), 283.
http://dx.doi.org/10.2307/2112876
Sewell, W., Haller, A., & Portes, A. (1969). The Educational and Early Occupational Attainment Process. American Sociological Review, 34(1), 82. http://dx.doi.org/10.2307/2092789
21 Sewell, W. & Hauser, R. (1975). Education, occupation, and earnings. New York: Academic Press. Sewell, W., & Shah, V. (1968). Social Class, Parental Encouragement, and Educational Aspirations.
American Journal Of Sociology, 73(5), 559-572. http://dx.doi.org/10.1086/224530 Stinebrickner, T., & Stinebrickner, R. (2012). Learning about Academic Ability and the College
Dropout Decision. Journal Of Labor Economics, 30(4), 707-748.
http://dx.doi.org/10.1086/666525
Van De Werfhorst, H., & Van Tubergen, F. (2007). Ethnicity, schooling, and merit in the Netherlands. Ethnicities, 7(3), 416-444. http://dx.doi.org/10.1177/1468796807080236
Woelfel, J., & Haller, A. (1971). Significant Others, The Self-Reflexive Act and the Attitude Formation Process. American Sociological Review, 36(1), 74. http://dx.doi.org/10.2307/2093508
Zafar, B. (2011). How Do College Students Form Expectations?. Journal Of Labor Economics, 29(2), 301-348. http://dx.doi.org/10.1086/658091
Zéroulou, Z. & Zeroulou, Z. (1988). La réussite scolaire des enfants d'immigrés: L'apport d'une approche en termes de mobilisation. Revue Française De Sociologie, 29(3), 447.
22 Appendix A: Weights
I used my own weights for this research. Schools are categorized into four strata. Stratum 1 contains schools with less than 10% students with a non-west background and the fourth strata consists of schools with more than 60% students with a non-west background. The table shows the number of schools in the Netherlands per stratum and the number of respondents in the sample per stratum. The weights are the number of schools divided by the number of respondents in a stratum.
Stratum Number of schools Respondents Weights
Stratum 1 (<0,1) 1116 465 2,400
Stratum 2 (0,1 - 0,3) 981 899 1,091
Stratum 3 (0,3 – 0,6) 247 731 0,338
23 Appendix B: Ethnic groups
The ethnic groups used in this research are based on the definition used in the CILS4EU. They distinguish 17 ethnic groups in the Netherlands:
Netherlands (natives) Turkey Morocco Suriname Indonesia Netherlands Antilles Germany Africa Latin America
Northern America and Oceania Southern Asia Western Asia Other Asia Southern Europe Western Europe Other Europe Unknown
Ethnic groups with too little respondents (N<50) such as Indonesia, Germany, Southern Europe and Western Europe, are not visible in the graphs.
24 Figures and tables
Figure 1 Schematic overview of the Dutch education system. Dutch names of tracks in parentheses.
Age Dutch education system
18 University Higher
vocational Lower vocational 17
Pre university (VWO) 16 Pre higher vocational (HAVO) 15 Pre lower vocational (high) (VMBO g/t) Pre lower vocational (low) (VMBO b/k) 14 13 12 11 Primary education 10 9 8 7 6 5 4
25 Figure 2 Aspirations parents and children. Y-axis represents current school type of the child. Data is weighted.
0 20 40 60 80 100
Pre low voc. (low) Pre low voc. (high) Pre high voc. Pre uni.
Children aspire university
Migrant Native
0 20 40 60 80 100
Pre low voc. (low) Pre low voc. (high) Pre high voc. Pre uni.
Children aspire higher voc.
Migrant Native
0 20 40 60 80 100
Pre low voc. (low) Pre low voc. (high) Pre high voc. Pre uni.
Parents aspire university
Migrant Native
0 20 40 60 80 100
Pre low voc, (low) Pre low voc, (high) Pre high voc, Pre uni,
Parents aspire higher voc.
26 Table 1 Analysis of missing values
Original N* N sample type 1** N Sample type 2*** Reduction sample 1 Reduction sample 2 Native 3013 2399 2003 -20,38% -33,52% Migrant 1097 747 384 -31,91% -65,00% Pre vocational secondary education (low) 1091 672 400 -38,41% -63,34% Pre vocational secondary education (high) 1401 1121 877 -19,99% -37,40% Senior general secondary education 831 699 542 -15,88% -34,78%
Pre university education 754 654 568 -13,26% -24,67%
Total 4077 3146 2387 -22,84% -41,45%
* Without western migrants
** Only those with missing values on student level variables removed and western migrants *** All respondents with missing values on student level variables and/or parent level variables and western migrants removed
27 Table 2 Descriptive statistics
Variable Mean SD Min Max N
Boy (=1) 0,488 . 0 1 3146
Migrant (=1) 0,138 . 0 1 3146
Current school type 1,560 1,040 0 3 3146
Grade 7,690 0,853 3 10 3146
SES 0,103 0,967 -1,960 1,870 3146
Educational aspirations children 5,370 1,700 0 7 3146
Educational aspirations parents 5,150 1,820 0 7 2387
Perceptions children 3,620 2,630 0 7 3146
Dummy educational aspirations children 1,220 0,710 0 2 3146
Dummy perceptions children 1,510 1,090 0 3 3146
Aspirations children Asp ira tio n s par ents
Pre low voc. Pre high voc. Pre uni. Low voc. High voc. Uni. Total
Don't know 12 17 4 10 30 42 115
10,43 14,78 3,48 8,7 26,09 36,52 100
Pre low voc. 48 40 4 18 20 8 138
34,78 28,99 2,9 13,04 14,49 5,8 100
Pre high voc. 13 72 8 13 68 33 207
6,28 34,78 3,86 6,28 32,85 15,94 100 Pre uni 0 5 7 3 15 43 73 0 6,85 9,59 4,11 20,55 58,9 100 Low voc 87 95 5 122 123 35 467 18,63 20,34 1,07 26,12 26,34 7,49 100 High voc. 23 146 35 48 366 230 848 2,71 17,22 4,13 5,66 43,16 27,12 100 Uni 3 17 24 5 70 420 539 0,56 3,15 4,45 0,93 12,99 77,92 100 Total 186 392 87 219 692 811 2,387 7,79 16,42 3,64 9,17 28,99 33,98 100
29 Table 4 Aspirations parents versus perceptions children (weighted)
Aspirations children Asp ira tio n s par ents Don’t know
Pre low voc. Pre high voc. Pre uni. Low voc. High voc. Uni. Total
Don't know 47 12 12 6 12 10 16 115
40,87 10,43 10,43 5,22 10,43 8,7 13,91 100
Pre low voc. 37 57 12 1 17 9 5 138
26,81 41,3 8,7 0,72 12,32 6,52 3,62 100
Pre high voc. 48 25 59 2 12 43 18 207
23,19 12,08 28,5 0,97 5,8 20,77 8,7 100 Pre uni 23 1 6 9 1 8 25 73 31,51 1,37 8,22 12,33 1,37 10,96 34,25 100 Low voc 111 122 40 4 129 45 16 467 23,77 26,12 8,57 0,86 27,62 9,64 3,43 100 High voc. 212 44 153 28 82 246 83 848 25 5,19 18,04 3,3 9,67 29,01 9,79 100 Uni 124 6 22 33 8 80 266 539 23,01 1,11 4,08 6,12 1,48 14,84 49,35 100 Total 602 267 304 83 261 441 429 2 387 25,22 11,19 12,74 3,48 10,93 18,48 17,97 100
30
Table 5 Logistic regression of children's ambition (weighted)
Model 1 Model 2 Model 3 Model 4 Model 5
Lower vocational (base category) Higher vocational
Migrant (=1) 0,767** 1,059*** 0,799* 0,664* 0,515
(0,286) (0,301) (0,313) (0,330) (0,339)
Boy (=1) -0,340* -0,168 -0,188 -0,279 -0,289
(0,135) (0,151) (0,155) (0,162) (0,163) Long term ambitions children (=1) 0,275* -0,138 -0,169 -0,102 -0,144
(0,138) (0,155) (0,162) (0,167) (0,172)
SES 0,464*** 0,274*** 0,218* 0,238** 0,200*
(0,0733) (0,0817) (0,0855) (0,0873) (0,0900)
Schooltype (ref. pre lower vocational (low)) . . .
Pre lower vocational (high) 1,263*** 1,023*** 0,983*** 0,859***
(0,168) (0,176) (0,179) (0,185)
Pre higher vocational 5,513*** 4,931*** 4,531*** 4,226***
(0,736) (0,743) (0,743) (0,750)
Pre university 18,14 16,85 15,95 16,18
(1084,9) (768,0) (481,1) (653,2)
Grades 0,253** 0,199* 0,232* 0,201*
(0,0888) (0,0912) (0,0951) (0,0967)
Parents ambitions (ref. lower vocational) . . .
Higher vocational 0,966*** 0,579** (0,178) (0,191) University 1,069* 0,487 (0,449) (0,476) Don’t know 0,222 0,116 (0,361) (0,390)
Long term ambitions parents (=1) -0,0460 0,110
(0,197) (0,210)
Perception children (ref. Lower vocational) .
Higher vocational 2,300*** 2,127*** (0,259) (0,266) University 1,364** 1,403** (0,478) (0,484) Don’t know 0,670*** 0,617** (0,187) (0,191) Constant 0,960*** -2,241** -2,001** -2,502*** -2,410** (0,128) (0,690) (0,727) (0,745) (0,779)
31
Table 5 Continuation
Model 1 Model 2 Model 3 Model 4 Model 5
University
Migrant (=1) 1,243*** 1,852*** 1,088** 0,996** 0,496
(0,290) (0,328) (0,352) (0,371) (0,389)
Boy (=1) -0,319* 0,198 0,222 0,0793 0,0966
(0,145) (0,186) (0,194) (0,200) (0,204) Long term ambitions children (=1) 1,683*** 0,607** 0,501* 0,592** 0,505*
(0,171) (0,209) (0,218) (0,225) (0,231)
SES 0,790*** 0,441*** 0,399*** 0,389*** 0,374***
(0,0793) (0,100) (0,106) (0,108) (0,112)
Schooltype (ref. pre lower vocational (low)) . . .
Pre lower vocational (high) 0,511* 0,171 0,441 0,198
(0,246) (0,263) (0,271) (0,281)
Pre higher vocational 5,723*** 4,848*** 4,793*** 4,234***
(0,753) (0,766) (0,766) (0,777)
Pre university 21,21 19,23 18,57 18,29
(1084,9) (768,0) (481,1) (653,2)
Grades 0,407*** 0,345** 0,425*** 0,379**
(0,110) (0,114) (0,119) (0,122)
Parents ambitions (ref. lower vocational) . .
Higher vocational 1,221*** 0,853** (0,265) (0,279) University 3,003*** 2,104*** (0,477) (0,507) Don’t know 0,811 0,627 (0,477) (0,512)
Long term ambitions parents (=1) -0,0828 0,0499
(0,254) (0,267)
Perception children (ref. Lower vocational) .
Higher vocational 2,423*** 2,141*** (0,343) (0,352) University 3,645*** 3,295*** (0,505) (0,520) Don’t know 1,301*** 1,102*** (0,287) (0,296) Constant -0,470** -5,217*** -5,096*** -6,297*** -6,123*** (0,166) (0,881) (0,941) (0,977) (1,023) Observations 2387 2387 2387 2387 2387 Pseudo R2 0,077 0,350 0,387 0,414 0,432
Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
32
Table 6 Sensitivity analysis. Logistic regression of children’s ambitions
Model 1 Model 2 Model 4
Sample Full Reduced Full Reduced Full Reduced
Lower vocational (base category) Higher vocational
Migrant (=1) 0,689** 0,767** 1,093*** 1,059*** 0,753** 0,664*
(0,229) (0,286) (0,243) (0,301) (0,262) (0,330)
Boy (=1) -0,308* -0,340* -0,192 -0,168 -0,269 -0,279
(0,123) (0,135) (0,136) (0,151) (0,144) (0,162) Long term ambitions children (=1) 0,310* 0,275* -0,102 -0,138 -0,0423 -0,102
(0,125) (0,138) (0,140) (0,155) (0,149) (0,167)
SES 0,422*** 0,464*** 0,196** 0,274*** 0,143 0,238**
(0,0663) (0,0733) (0,0742) (0,0817) (0,0787) (0,0873) Schooltype (ref. pre lower voc.
(low))
. .
Pre lower vocational (high) 1,399*** 1,263*** 1,200*** 0,983***
(0,149) (0,168) (0,158) (0,179)
Pre higher vocational 5,684*** 5,513*** 4,790*** 4,531***
(0,732) (0,736) (0,738) (0,743)
Pre university 4,394*** 18,14 3,965*** 15,95
(1,117) (1084,9) (1,121) (481,1)
Grades 0,227** 0,253** 0,180* 0,232*
(0,0781) (0,0888) (0,0827) (0,0951)
Perception children (ref. lower voc.) .
Higher vocational 2,016*** 2,077*** (0,212) (0,223) University 0,968** 0,881* (0,369) (0,376) Don’t know 0,664*** 0,621*** (0,164) (0,168) Constant 0,947*** (0,114) 0,960*** (0,128) -2,075*** (0,608) -2,241** (0,690) -2,077*** (0,627) -2,190*** (0,647)
33
Table 6 continuation sensitivity analysis
Model 1 Model 2 Model 4
Full Reduced Full Reduced Full Reduced
University
Migrant (=1) 1,303*** 1,243*** 2,030*** 1,852*** 1,276*** 0,996**
(0,230) (0,290) (0,257) (0,328) (0,284) (0,371)
Boy (=1) -0,253 -0,319* 0,194 0,198 0,117 0,0793
(0,132) (0,145) (0,164) (0,186) (0,175) (0,200) Long term ambitions children (=1) 1,821*** 1,683*** 0,840*** 0,607** 0,837*** 0,592**
(0,156) (0,171) (0,185) (0,209) (0,196) (0,225)
SES 0,699*** 0,790*** 0,283** 0,441*** 0,203* 0,389***
(0,0715) (0,0793) (0,0879) (0,100) (0,0941) (0,108)
Schooltype (ref. pre lower voc. (low)) . . .
Pre lower vocational (high) 0,395* 0,511* 0,345 0,441
(0,201) (0,246) (0,217) (0,271)
Pre higher vocational 5,591*** 5,723*** 4,698*** 4,793***
(0,741) (0,753) (0,749) (0,766)
Pre university 7,235*** 21,21 6,258*** 18,57
(1,113) (1084,9) (1,117) (481,1)
Grades 0,385*** 0,407*** 0,354*** 0,425***
(0,0946) (0,110) (0,102) (0,119) Perception children (ref. lower
vocational) Higher vocational 2,283*** 2,423*** (0,294) (0,343) University 3,190*** 3,645*** (0,394) (0,505) Don’t know 1,348*** 1,301*** (0,248) (0,287) Constant -0,575*** -0,470** -4,920*** -5,217*** -5,694*** -6,297*** (0,150) (0,166) (0,758) (0,881) (0,832) (0,977) Observations 3146 2387 3146 2387 3146 2387 Pseudo R2 0,082 0,077 0,335 0,350 0,394 0,432
Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001
