Educational Mobility of Second-generation Turks

Philipp Schnell onderzocht wat de verschillen veroorzaakt in de onderwijsmobiliteit van tweede generatie Turken in vijf steden in Zweden, Frankrijk en Oostenrijk. Hij deed onderzoek in Stockholm, Parijs, Straatsburg, Wenen en Linz. Hij keek naar factoren op individueel niveau, zoals familieleden die hulpbronnen inbrachten relevant voor de onderwijsmobiliteit. Schnell bestudeerde ook algemene karakteristieken van drie onderwijssystemen en hun onderlinge verschillen. Hij concludeert dat twee elementen betrokken zijn bij het mobiliteitsproces: kinderen van Turkse immigranten (met hun eigen individuele kenmerken, gezinsachtergronden en relaties met belangrijke tussenpersonen zoals peers en leraren) en de onderwijssystemen van de landen (elk met hun verschillende institutionele regelingen). Het is de interactie tussen deze elementen die de richting en de uiteindelijke uitkomst van de mobiliteit in het onderwijs bepaalt. De onderwijssystemen, die institutionele regelingen kennen die de onderwijsmobiliteit van tweede generatie Turken minder afhankelijk maken van factoren en hulpbronnen op individueel niveau, leiden tot betere onderwijsprestaties. Het onderzoek naar deze interacties gedurende de gehele schoolcarriere van een leerling toont de gunstige en minder gunstige institutionele omstandigheden die van belang zijn in het begrijpen van verschillen in onderwijsmobiliteit in een internationale vergelijking.


Introduction
This chapter provides an initial glance at the education outcomes at the aggregated level in and across Austria, France and Sweden.The aim is to draw an initial picture of how the Turkish second generation is positioned in terms of their education and how mobile they are based on the empirical material.Three perspectives are included: Firstly, the levels of education attained by the Turkish second generation will be explored in relation to the respective comparison groups by asking whether equality of outcomes can be observed between the two groups in each individual country.Special attention will be given to parents' education backgrounds in explaining potential differences in education outcomes between the two compared groups.
Secondly, the levels of education attained by the Turkish second generation in each of the three countries will be compared.Variations in the outcomes of the second generation will be examined to see whether they are attributable to compositional differences in the Turkish parental generation (see chapter 2).The aim of this procedure is to explore whether children of Turkish immigrants who come from similar family homes achieve similar outcomes growing up in different countries and cities.
Finally, the extent to which the Turkish second generation demonstrates inter-generational progress is included as a third perspective on outcomes.This will be done by exploring the degree to which second-generation Turks move beyond the levels of education reached by their parents.
All three perspectives outline the position of second-generation Turks -in terms of their levels of education -in a comparative way.This first glimpse of education outcomes with these three comparison perspectives not only shows the education mobility processes of second-generation Turks, but also raises the question of whether the formation of patterns of ethnic inequality become evident across countries (D'Addio 2007;Rumbaut 2008).

Education levels reached by young adults in Austria, France and Sweden
This section explores the education levels reached by second-generation Turks in relation to the comparison group in each city and country by focusing on whether there is equality of outcomes between the two groups.
In the sociology of education, 'educational positioning' is usually assessed by using the highest diploma of the target population (the highest level of education attained) as the reference for comparisons across countries.However, as mentioned earlier in the introduction to this book, many of the Turkish second generation we interviewed were still enrolled in school. 1 This is especially true of second-generation Turks in France, where more than every second respondent was still enrolled in some sort of education at the time they were interviewed.So following the 'traditional' measurement of educational attainment by looking at their highest diploma would blur the picture and underestimate the current situation.The dependent variable for this chapter is therefore the education level of the respondents.This is defined as a combination of the highest obtained diploma for those who have already left school and the current level of education for those still in school.This allows a clearer comparison to be made between the second generation in Austria, France and Sweden.

How school outcomes differ between the groups
Table 3.1 shows the distribution of education levels for second-generation Turks and their comparison group in all five cities.Education levels are displayed using the 'Edu codes' classification scheme, which ranges from the lowest (primary education) to the highest (tertiary education) in each country (see Crul, Schnell, Herzog-Punzenberger, Wilmes, Slootman & Aparicio-Gomez 2012, and Appendix B table A9).Turning to the results for Austria, the Turkish second generation more frequently leaves school with only a compulsory education certificate (primary and lower-secondary education together) compared to the comparison group.About one-third of the Turkish second generation in Austria obtains an apprenticeship certificate or similar as their highest education level.Moreover, they are significantly over-represented in the vocational track, with a difference of 10 per cent between them and the comparison group in Vienna and a 15 per cent difference in Linz.In addition to these general national trends, city-to-city differences can be seen in Austria as well.Second-generation Turks in Linz are less represented in the academic track in upper-secondary education but more often found in the higher vocational track.Apart from these differences, the overall educational distribution of second-generation Turks between Linz and Vienna shows similar outcomes, while the chi-squared test (see bottom of table 3.1) indicates significant variations between the comparison groups in the two cities.The latter is most evident when considering the proportion of students leaving school at the end of their compulsory education.In Vienna, 14 per cent do not continue after lower-secondary education, compared with 8.6 per cent in Linz.

Austria
The results for the French cities are displayed in the middle columns of table 3.1.Overall, the descriptive outcomes highlight three major findings.Firstly, the overall education levels of each group, and therefore the general student population, is far higher in France than in Austria.This countryspecific feature is in line with official statistics on the distribution of educational attainment (Eurostat 2010).In the mid-1980s, education reforms were introduced in France in order to increase the number of people holding the baccalauréat, or 'bacc' (Duru-Bellat & Kieffer 2001;Kieffer 2008).In uppersecondary education, higher vocational and academically orientated schools (lycées) enable students to obtain the 'bacc'.Recent studies have shown that these reforms made the 'bacc' the education norm in France, and that the requirements of the labour market have shifted into the post-secondary and tertiary education sector (Brauns, Steinmann, Kieffer & Marry 1999;Duru-Bellat & Kieffer 2008;Duru-Bellat, Kieffer & Fournier-Maerelli 1997).Consequently, the French education system has moved to a mass system of post-secondary and tertiary education over the past three decades.
Secondly, figures for France indicate that Turkish second-generation students are more often at the lower end of the education spectrum than their peers in the comparison group, and they fall behind in post-secondary and tertiary education.They leave school after compulsory education (primary and lower-secondary education) in higher numbers, while the difference in post-secondary and tertiary education amounts to around 26 per cent when both cities are taken together.This gap between the two groups in post-secondary and tertiary education mirrors previous findings for the educational attainment of second-generation Turks in France, which showed a gap of almost the same size (Kirszbaum et al. 2009: 26).In addition, the overall distribution of education levels presented in table 3.1 also resembles the results for the educational attainment of second-generation Turks from a recent survey of children of immigrants in France (Brinbaum, Moguérou & Primon 2010).
Finally, variations in the distribution of education levels can be seen between second-generation Turks in Strasbourg and Paris.In Strasbourg, children of Turkish immigrants are more likely to attend vocational schools (for apprenticeships and similar), and consequently advance less frequently beyond upper-secondary education.When considering post-secondary and tertiary education, the Turkish second generation in Paris achieves this level almost twice as often as their counterparts in Strasbourg.
The last two columns on the right side of table 3.1 show the results for Stockholm.At a first glance, the overall distribution of the education levels resembles that of France.The Swedish education system also consists of a comprehensive system and a post-secondary and tertiary sector in which job specialisation takes place.Consequently, the great majority of students are situated on the upper end of the education ladder.Nevertheless, significant group differences in education levels are recognizable.Second-generation Turks are twice as likely as comparison group students to leave Swedish schools after compulsory education.This finding is in line with the results for the educational attainment of children of Turkish immigrants in Sweden (Jonsson & Rudolphi 2008).Moving on to the top end of the education ladder, the gap between the two compared groups amounts to nearly 30 per cent in favour of the comparison group.

Associations with the parents' levels of education
How can the differences in education outcomes reported above between the Turkish second generation and their comparison group be explained?Of course, a detailed determination of multiple potential explanatory factors goes beyond the scope of this first overview and frames the tasks of the forthcoming chapters.However, this section considers one of the most important factors that might explain quite a few of the differences: the parents' levels of education.
The link between social origin (including level of education) and the educational attainment of children has been identified as the major explanation for the disadvantaged position of the children of immigrants -including the Turkish second generation -in education in Europe (Alba, Handl & Müller 1994;Crul & Holdaway 2009;Heath & Brinbaum 2007;Heath et al. 2008;Kalter, Granato & Kristen 2007;Rothon et al. 2009;Vallet & Caille 1996).Previous studies demonstrate that the Turkish second generation in Europe often comes from less-advantaged social and education backgrounds (Crul & Vermeulen 2003;Crul et al. 2012;Dustmann et al. 2012, compare also chapter 2).It is therefore possible that many of the differences between the comparison group and second-generation Turks can be explained by differences in their parents' education backgrounds. 2he measurement of parental education is based on the parent with the highest qualif ications and consists of f ive categories: primary or below, lower-secondary, upper-secondary, post-secondary, and tertiary education.The distribution of the parental education levels is displayed in table 3.2.The first point to note is that the education levels of the Turkish parents is significantly lower than that of the comparison group across countries.As indicated in chapter 2, the great majority of Turkish fathers and mothers migrated at an early age for work reasons, and were equipped with little or no experience of education.The proportion of less-educated parents (primary at best) is highest among Turkish fathers and mothers living in Strasbourg.Stockholm has the highest share of well-educated parents (compare chapter 2).This is partially the result of additional training programmes in Sweden and the substantial group of Turkish refugees who obtained higher education credentials when they arrived.Overall, second-generation Turks living in all three countries come from families with lower levels of education.It is also worth noting that the education levels of parents in the comparison group vary from country to country.Sweden and France show higher proportions of post-secondary (or higher) diploma-holders, while the majority of fathers and mothers in the Austrian comparison group mostly achieve upper-secondary level.The next step in this book will be to check for statistical differences between groups, showing each individual respondent's education background.This will include an examination of how many of these differences can be accounted for by differences in levels of education.Given the diverse overall education profiles of the student population across the three countries, the education outcomes of the respondents (dependent variable) will be reduced further and standardised into four broad categories: (1) primary and lower-secondary (compulsory education), (2) apprenticeships and similar, (3) upper-secondary, and (4) tertiary education.Gaps in education will be presented in odds ratios, which are a useful measure for assessing relative performance since they are independent of the overall achievement (Rothon et al. 2009(Rothon et al. : 1422)).
The advantage that's gained by using a relative measure such as odds ratios rather than a conventional measure (such as the percentage complet-ing a certain level of education) is important since it is not the absolute level of education that matters but whether second-generation Turks have more or fewer opportunities than the comparison group with whom they compete in schools, and, later on, in the labour market of each city.Having highly educated parents might be an advantage in countries such as Austria where fewer pupils have more highly educated backgrounds, whereas in countries such as France the overall level of educational attainment is higher and more frequently distributed in the population.
To facilitate interpretation, any gaps in education between groups are illustrated in figure 3.1.Values above 1 indicate better chances of achieving the education level in question among second-generation Turks than the comparison group.Values below one indicate fewer chances.An exact value of one for both groups would be a sign of equal opportunity.In short, the higher any bar is from 1, the bigger the education gap will be between the two compared groups.Results are presented before ('gross': black bars) and after checking for parents' levels of education ('net': grey bars).The aim of this strategy is twofold: to examine the education gaps independent of the overall achievement, and to explore whether these gaps are biggest at the lowest and highest ends of the education spectrum in all three countries (black bars).In all three countries, second-generation Turks are, on average, more than two and a half times more likely to finish school with, at best, a certificate from compulsory school.And, at the top end of the education spectrum, the chances of completing their education with a post-secondary or tertiary qualification are (on average) only half as big for the Turkish second generation as for the comparison group.In line with the descriptive findings in table 3.1, the greatest variety between cities is found in uppersecondary education.In the two French cities and in Vienna, significant differences between the compared groups are absent, while the Turkish second generation in Stockholm significantly outperforms the comparison group.Do these differences in education levels persist once the parental education level is 'held constant'?The grey bars show the odds ratios for second-generation Turks relative to the comparison group after checking for parents' levels of education.One conclusion emerges clearly from the grey bars displayed in figure 3.1: the differences in levels of education between the two groups can, to a great extent, be explained by their parents' levels of education.Especially at the lower end of the spectrum, significant differences mostly vanish after checking for parents' levels of education.This pattern of explaining the education gap in the lower education categories appears to be common in all cities except Strasbourg, where the gap persists significantly.
It is also worth noting that in post-secondary and tertiary education, overrepresentation of the comparison group remains significant in Stockholm, Strasbourg and Vienna.The differences between the odds ratios before and after checking for parents' levels of education do not decrease substantially at the upper end of the education spectrum.Considering the 'ranking' of the five cities, the patterns do not change much between the gross and net differences.Overall, differences in education levels seem to be smallest in Paris and Stockholm on the one hand and biggest in the Austrian cities on the other hand, while Strasbourg is located in between.The results in figure 3.1 provide a first glimpse of the size and strength of the influence of parents' education in the three countries.Parents' education levels account for most of the differences observed in the Austrian and French cities (compare black and grey bars), while in Sweden they account for slightly less.

Accounting for compositional differences
Understanding and interpreting the differences in education outcomes for the Turkish second generation across countries is a complex matter.Why are children of Turkish immigrants predominantly achieving lower levels of education in Austria compared to their age-mates of Turkish origin in France and Sweden (compare table 3.1 and figure 3.1)?'Origins shape destinies' writes Rumbaut (1999: 187) and hints at the contribution that compositional differences in the first generation can make in explaining education disparities among the second generation.These differences in education and occupational composition are often related to immigrant selectivity (Feliciano 2005a(Feliciano , 2005b)), and this line of argument is frequently applied when comparing the outcomes of children of diverse ethnic origin groups in one single country (e.g.Haller, Portes & Lynch 2011;Heath et al. 2008;Portes & Rumbaut 2001).Although the methodological approach applied in this study is different in that it compares children of the same origin group (Turks) across countries, compositional differences may play a role as well.The great majority of Turkish fathers in all three countries migrated for work reasons while mothers overwhelmingly followed to reunify families, and both parents frequently had few educational qualifications at their disposal.Nevertheless, some degree of heterogeneity within the Turkish communities has been observed as well in chapter 2especially across the three compared countries.Variations have been found in terms of the reasons for migration, time of arrival and regions of origin.
Compositional differences are related not only to the migration histories of Turkish fathers and mothers, but were shaped and reinforced by the conditions immigrants faced upon arrival in their receiving countries and cities.Reitz (1998Reitz ( , 2002;;Reitz & Somerville 2004) has called the consequences of these varying conditions on children of immigrants 'indirect effects of host society institutions'.Host society institutions, such as national and local labour markets, may affect the Turkish second generation indirectly through their parents' experiences, and may serve as an explanation for some of the differences in their outcomes across countries.For example, Turkish fathers and mothers in Stockholm have been found to obtain, on average, higher-prestige jobs and more host-country education credentials than Turkish immigrants in France and Austria -which may later translate into better financial circumstances, more educationally relevant resources and a deeper knowledge of the workings of the national education system.In other words, the varying range of conditions Turkish families faced on arrival may have affected the socio-economic position of the first generation and therefore 'indirectly' the family lives and circumstances in which the Turkish second generation grew up.
This section explores whether the differences in the education outcomes of second-generation Turks across cities remain significant once those outcomes are adjusted for compositional differences in the Turkish first generation.Only if significant education disparities exist after controlling for these family-related characteristics is further analysis justified in order to assess the direct impact of other relevant factors which may explain cross-national differences.
Table 3.3 shows the results of a multivariate analysis (ordered logistic regression).The dependent variable is the education level of the Turkish second generation, coded as lower-secondary at the most (1), upper-secondary and apprenticeship (2), and post-secondary/tertiary education (3).Model 1 displays the differences in education outcomes among second-generation Turks across cities expressed in odds ratios.The estimates in Model 1 have already been tested for the socio-demographic characteristics (age and gender) of the second generation and whether or not the respondents are still enrolled in school.The results of Model 1 resemble the descriptive patterns observed in table 3.1.The odds of achieving a higher level of education are around three and four times higher in Stockholm and Paris respectively than in Vienna.The odds of second-generation Turks in Strasbourg achieving a higher level of education are lower than in Paris or Stockholm, but still significantly higher than for second-generation Turks in Vienna.The non-significant coefficient for Linz indicates similar opportunities for second-generation Turks in both Austrian cities.In a second step (Model 2), a number of independent variables has been introduced in order to explore whether differences in outcomes between the Turkish second generation across cities can be related to composition differences in the parents' generation.These independent variables are: the education levels of the parents, whether at least one parent attended school in the survey country, whether both parents were employed when the respondent was 15, family size and parents' language ability in German, French or Swedish.The differences we have already observed between the reference group, second-generation Turks in Vienna, and second-generation Turks in Stockholm, are slightly reduced, while they remain identical in the French cities.In other words, even after testing for compositional differences in the first generation, variations in the outcomes for their children across countries and cities remain significant.Finally, migration-related variables for the parents have been added to the model.These variables are also taken from chapter 2 and include the length of time parents have resided in the survey city, whether the parents originate from a highly developed area of Turkey, and whether or not fathers migrated primarily for reasons of work.The advantaged position of the Turkish second generation in Paris and Stockholm is slightly reduced when holding these variables constant, but the overall differences in attempts to achieve a higher level of education do not disappear.
The main conclusion to be drawn from table 3.3 is that second-generation Turks in France and Sweden are still significantly more likely to achieve a higher level of education than the second generation in Austria, even after testing statistically for compositional differences in the parental generation.Further, Reitz (1998Reitz ( , 2002) ) has argued that country institutions exert an indirect effect on the position of the second generation, and this may explain some of the cross-national variations.The analysis presented in this section indicates, however, that after applying controls, differences in the education outcomes of the Turkish second generation remain highly significant across cities and countries.

Inter-generational educational mobility
So far, the education outcomes of second-generation Turks have been examined in relation to the comparison group (Section 3.2) and across countries (Section 3.3).The last comparative perspective of this first glimpse at education outcomes investigates inter-generational mobility between the Turkish first and second generations.A large body of work has shown that opportunities for children of immigrants become most evident by examining their degree of inter-generational mobility (De Broucker & Underwood 1998;Loury 2005;Portes & Rumbaut 2001;Platt 2005;Rumbaut 2004Rumbaut , 2008)).Greater levels of inter-generational mobility can be read as indicators of greater openness and a weaker link between the parents' levels of education and advantageous outcomes for their children.Further, the inter-generational mobility approach also indicates whether or not children of Turkish im-migrants improve their educational distribution.This section builds on the work cited above by asking how patterns of educational attainment among young adults of Turkish origin compare to those of their parents.In order to assess whether the Turkish second generation made some progress in relation to the first generation, a mobility index has been created.Following the procedures applied by Rumbaut ( 2008) to measure inter-generational mobility, the education outcomes of the second generation have been coded into the same categories as the parents' education levels (see table 3.2).Second-generation offspring who obtained the same levels of education as their parents are labelled 'immobile', while those found to be either above or below their parents' levels are classified as 'upward movers' or 'downward movers' respectively. 3From this perspective, the general trend of mobility between the generations is of interest, rather than looking only at actual achieved levels of education.Parents of mixed parentage and those who achieved their highest educational credentials in the schools of the host country have been excluded from the analysis in order to avoid distortions.
The results of the mobility patterns are given in table 3.4.The data demands make it unfeasible to apply them within this study, given the relatively small N. Within this section, results on the absolute rather than the relative degree of mobility are presented by using descriptive measures of inter-generational mobility.
At first glance, a predominantly upward trend in education mobility from the first generation to the second can be seen in all three countries.To begin with, in Sweden almost nine out of ten children of Turkish origin achieve a higher level of education than their parents, while only about 3 per cent perform below the level of their parents.In France, more than 70 per cent are upward movers.The picture in the Austrian cities is more diverse.There, only slightly more than one child in two moves beyond the level of his parents, and almost every third child achieves a level of education similar to that of his parents.The number of downward movers is also highest in Austria with around 14 per cent in each of the cities.It is worth noting that the results from the two French cities, Paris and Strasbourg, differ in terms of both downward movers and those who remained educationally immobile.In Strasbourg, more second-generation Turks move beyond or achieve the same level as their parents when compared to their counterparts in Paris.The lower levels of downward mobility are not surprising given the relatively low education level of the Turkish first generation in Strasbourg (see table 3.2 and compare with chapter 2).Taken as a whole, the findings of the inter-generational educational mobility index point in a similar direction to the results presented in the previous sections.In Austria, the educational opportunities for second-generation Turks seem to be blocked in comparison with either France or Sweden, suggesting greater meritocracies in the latter.These results become even more pronounced when investigating 'long-range upward mobility'.These are upward movers who leap over at least one education category (Breen 2004b).For example, second-generation Turks whose parents had at least a lower-secondary level of education, and who achieved post-secondary education themselves, would be labelled as 'long-range upward movers'.Figure 3.2 displays the distribution of long-range upward mobility within the three countries and five cities.Not surprisingly, the overall mobility trend observed in table 3.4 remains in figure 3.2 as well: long-range upward mobility is two to three times more common in France and Sweden than in Austria.Around 30 per cent of students from the Turkish second generation in Paris and Stockholm move at least two levels beyond their parents.The average rate in the Austrian cities is around 10 per cent, while second-generation Turks in Strasbourg are in between, comprising around 20 per cent long-range upward movers.This latter finding underscores especially the dissimilar patterns between the French cities.The Turkish second generation in Strasbourg has had a predominantly upward journey because the educational attainment of the first generation was particularly low.At the same time, the number of children who climb far higher than the level of their parents is lower in Strasbourg than in Paris.

Conclusion
This chapter provided a first glimpse of education outcomes at the aggregated level by applying three levels of comparison: the Turkish second generation with the comparison group within countries, with secondgeneration Turks across countries, and between generations.
To begin with the latter, there has been substantial inter-generational progress by the Turkish second generation as measured descriptively relative to the parental generation.In comparison with their parents, the overwhelming majority of the children showed some progress in the education sphere, but at very different rates.The Turkish second generation in Sweden, and to a lesser extent in France, displayed exceptional achievement.By comparison, the achievements of the second generation in Austria were moderate.So, the predominantly upward trend in mobility in relation to the parental generation across the countries is a success story, but it should also be regarded with caution because the majority of Turkish parents migrated to these various countries equipped with very little experience of education.
Turning to the relative comparison within countries and cities, the findings indicate significant differences in the distribution of education levels between the compared groups.On average, the differences between the comparison group and second-generation Turks are most pronounced at the lowest and the highest ends of the education ladder in all three countries.To a substantial degree, the disadvantaged position of second-generation Turks within countries can be related to parents' levels of education since children of Turkish origin are more likely to come from less-educated families.This makes it difficult to climb the education ladder.Nevertheless, significant disadvantages did not always vanish after testing statistically for parents' levels of education.
This first glance at the outcomes provided a third perspective by comparing the education levels of second-generation Turks across cities.The fact that second-generation Turks in France and Sweden outperform their counterparts in Austria could not be explained by the compositional differences in the first generation, which were detected in chapter 2.Even after testing for a wide range of composition-related and migration-related factors in the parental generation, the odds of the second generation achieving a similar education level in Vienna and Linz were still significantly lower than in either of the French cities or in Stockholm. 4hese provisional results raise a number of questions that will pave the way for the forthcoming chapters.The first question concerns the poorer achievements of second-generation Turks in relation to the comparison groups in all three countries.Although parents' levels of education played a substantial role in explaining parts of the disadvantaged position of the Turkish second generation, the specific mechanisms through which parents' education correlates with children's learning and education outcomes still remain elusive.Other individual and group-related factors that affect the schooling outcomes of second-generation Turks should be considered.For example, more information is needed about the availability of educational resources in families, as well as outside the family home.In addition to these individual-level explanations, there appears to be a number of institutional factors at play as well.In order to investigate, education pathways have to be examined.Focusing on school careers will provide detailed insights into the mechanisms at work between the institutional arrangements of schools and individual characteristics at different stages.The following chapters

Figure
Figure 3.1 Differences in education outcomes before ('gross') and after ('net') controlling for parents' education levels, by city (odds ratios)

Figure
Figure 3.2 Long-range upward mobility in second-generation Turks, by city (%)

Table 3 .1 Education levels of second-generation Turks and the comparison group, by city (%)
Turning to the upper end of the education spectrum, post-secondary and tertiary education, the figures indicate that the comparison group is clearly out-performing second-generation Turks in both Austrian cities.

Table 3 .3 Ordered logistic regression of education levels (odds ratios)
Notes: ordered logistic regression on educational level of the second generation.dependent variable: lower secondary at the most (1), upper-secondary and apprenticeship (2) and post-secondary/ tertiary education (3).Model specifications and distribution of the independent variables are given in appendix b. levels of significance:* p<0.05, ** p<0.01, *** p<0.001.n.s.=not significant.

Table 3 .4 Inter-generational educational mobility of the Turkish second generation, by city (%)
These multivariate mobility tables have been developed for analysing large-scale surveys.