From the Baccalauréat to Higher Education in France: Shifting Inequalities

Abstract

The extent to which education in France has been democratized is subject to debate: although school enrolment rates have continued to rise, does this trend reflect a true reduction in the social inequality of access to the various levels and courses of school and university education? Following an article published in Population in 2000, whose title evoked this controversial issue, Marie Duru-Bellat and Annick Kieffer conduct a detailed analysis of the mechanisms by which inequalities are reproduced in higher education in France. Their results bring clear critical evidence to the debate by showing that the quantitative opening-up of the education system is still accompanied by a marked social differentiation of students enrolled in different courses and subjects.

INSEE's most recent survey in 2003 makes it possible to evaluate changes in higher education in two birth cohort groups at either end of the period of rapid expansion in access to the baccalauréat (upper secondary exit examination) from 1985 to 1995. We examine how this opening up affected social inequalities in access to and success in higher education. The undeniable democratization of the baccalauréat has been followed by a more limited democratization in access to higher education. The first wave concentrated new working-class students in vocational baccalauréats; but owing to the strong links between secondary and higher education in France, these students' options in higher education have been restricted as a consequence. In particular, access to the selective elite grandes écoles has seen no democratization among holders of the baccalauréat, whereas shorter vocational courses and non-selective university courses have opened up. This limits the effect on social mobility of democratization of education at this level, because career opportunities increasingly depend not so much on the level of qualification but on the subjects chosen.

Résumé

La derniére enquête FQP réalisée par l'Insee en 2003 permet d'évaluer les transformations intervenues au niveau de l'enseignement supérieur dans deux groupes de cohortes encadrant la période d'expansion très forte de l'accès au baccalauréat entre 1985 et 1995. La question posée est de savoir comment cette ouverture a affecté les inégalités sociales d'accès et de réussite dans l'enseignement supérieur. L'indéniable démocratisation du baccalauréat se traduit par une démocratisation plus limitée de l'accès au supérieur. En effet, la première vague de démocratisation s'est traduite par une concentration accrue des nouveaux bacheliers de milieu populaire dans les baccalauréats professionnels ; compte tenu de la force des articulations entre enseignement secondaire et supérieur en France, ce phénomène a induit une limitation des possibilités d'orientation dans l'enseignement supérieur. En particulier, l'accès aux filières d'élite n'a connu aucune démocratisation parmi les bacheliers, alors que les filières professionnelles courtes et les filières universitaires s'ouvraient. Ceci limite la portée de la démocratisation de l'enseignement è ce niveau en termes de mobilité sociale, puisque de plus en plus, les débouchés ne dépendent pas du niveau de diplôme, mais du type de filière suivie.

Resumen

La última encuesta FQP realizada por el Insee en 2003 permite evaluar las transformaciones que han tenido lugar a nivel de la enseñanza superior en dos grupos de cohortes que enmarcan el período de expansión muy fuerte del acceso al bachillerato (baccalauréat) entre 1985 y 1995. La pregunta que se plantea es la de saber cómo esta apertura ha influido en las desigualdades sociales de acceso y de éxito en la enseñanza superior. La innegable democratización del bachillerato se traduce en una democratización más limitada del acceso a los estudios superiores. En efecto, la primera ola de democratización se reflejó en una concentración incrementada de los nuevos bachilleres de medio popular en los bachilleratos profesionales; teniendo en cuenta la fuerza de las articulaciones entre enseñanza secundaria y superior en Francia, este fenómeno ha inducido una limitación de las posibilidades de orientación en la enseñanza superior. En particular, el acceso a las carreras más selectas no ha experimentado democratización alguna entre los bachilleres, mientras que las carreras profesionales cortas y la carreras universitarias se abrían. Ello limita el alcance de la democratización de la enseñanza a este nivel en términos de movilidad social, puesto que cada vez más, las salidas y oportunidades no dependen del nivel de diploma sino del tipo de itinerario seguida.

The topic of the democratization of education has been extensively studied in French social science research. There is nonetheless a continuous need to update data and interpretations to take account of changes affecting the educational system over time. In this respect, the 1985-1995 period is of particular interest, since it saw a spectacular increase in the proportion of each cohort passing the baccalauréat examination, and thus achieving a qualification at the end of seven years in secondary education (between the ages of 17 and 19). The proportion of baccalauréat-holders per cohort, which was around 5% in 1950, rose from 20% in 1970 to 29.4% in 1985, 43.5% in 1990 and 62.7% in 1995. It has remained stationary since that date.

This order-of-magnitude change in the number of young people gaining a qualification at the end of secondary school occurred within an increasingly diversified system. Since the baccalauréat in France entitles the holder to enter higher education, the number of university students also rose sharply, with a doubling of the number of graduates from 1985 to 1996. Against this background of a rapid change that appears now to have levelled off, it is interesting to assess the forms that the democratization of higher education has taken and to examine how course choices have changed as a function of students' personal characteristics.

Higher education has grown considerably in all European countries, and the European Union authorities have stressed the need to develop a "knowledge society". This "quantitative" development has usually involved a diversification of structures. Intended to achieve a more efficient match between educational "supply" and demand from the more diverse groups who are to be educated, this diversification also gives rise to an increasingly hierarchical differentiation between courses, with its ensuing social implications. A leading issue in sociology in Europe is thus to examine whether the social inequalities in higher education (and in access to higher social status) are indeed undergoing radical change. Such inequalities, it is argued, concern more the access to particular courses than to a certain level of education; social differentiation between courses, in that case, being more important than differentiation in level. Indeed, "[t]he co-occurrence of expansion and differentiation is the basis for claims that higher education expansion is primarily a process of diversion, channelling members of the working class to lower-status post-secondary opportunities in order to reserve higher-status opportunities for the elite" (Arum et al., 2007, p. 5).

Recent analyses on this point are available for France (Merle, 1996, 2000 and 2002; Albouy and Wanecq, 2003; Selz and Vallet, 2006), but the considerable changes that have occurred in the various secondary and higher education courses in the last twenty years have not been sufficiently explored, particularly the matter of course choice in higher education. Given that one young person in two now leaves the education system at tertiary level, this aspect is essential for any assessment of how far and in what ways democratization has occurred. For this purpose we make use of Merle's typology (2000) to distinguish between equalizing democratization whereby gaps between groups are reduced, uniform democratization whereby the gaps are simply transferred into higher education, and segregative democratization, whereby inequalities work their way into the differentiation of types of study and speciality. A corollary question is the connection between inequality in the length of education and inequality in the type and specialization of that education.

Changes in the social selectiveness of higher education can only be addressed by examining the quantitative and qualitative changes in the eligible population, namely holders of the baccalauréat. This is presented in the first part, followed by an examination of the move from secondary to higher education, and career choices at this point, according to type and speciality of study. We use data from the latest vocational training and qualification survey (Formation et qualification professionnelle, FQP) carried out by INSEE in 2003⁽¹⁾.

I. Changes in access to the baccalauréat

A general context of democratization

As school enrolment has developed in France, there has been a clear process of democratization in access to secondary education, with a shift of inequality from the beginning of year 6 (6e in French numbering) to the beginning of year 10 (2nde) during the 1970s and 1980s⁽²⁾ (Duru-Bellat and Kieffer, 2001; see also Vallet, 2001). Only recently has access to the upper secondary level in year 10 (lycée) been opened up: no reduction in inequality of access to year 6 was observed between children of parents in higher-level occupations and manual workers' children from 1972 to 1980⁽³⁾, whereas a significant reduction occurred from 1980 to 1989. At the start of this period, the lycée enrolment rates were 90% for the children of persons in higher-level occupations and 42% for the children of manual workers, and any further expansion logically involved reducing the gap as the more advantaged groups reached a "ceiling". This is what Merle (2000) calls "equalizing democratization".

The development of secondary school enrolment was the result of an explicit policy announced in the early 1980s: "the baccalauréat for 80% of an age group". It involved reducing the number of pupils repeating years in primary school and abolishing early diversion into vocational education which, until the mid-1980s, meant that the weakest pupils – also most often working-class ones – left secondary school at the age of 14-15. This reduced the degree of selectivity in school careers, even if pupils' academic level did not significantly improve. Some authors even spoke of a relative under-selection of working-class pupils that was liable to affect their later educational trajectory (Convert, 2006). At all events, a more diverse population was entering the full-time secondary school system, and needing to make choices within a system that had become considerably more diversified. In particular, a new vocational baccalauréat was created in 1985 and awarded for the first time in 1987-1988 to young people who had been directed after year 9 into a two-year course in the vocational section of secondary school⁽⁴⁾. It was clear that the objective of the baccalauréat for 80% of the age group could not be achieved unless the system accepted pupils of lower academic attainment and more working-class background⁽⁵⁾.

As Table 1 shows, the rate of access to the general baccalauréat has not risen since 1995 although it had doubled since 1970. Although the rise was a sharp one for all types of baccalauréat from 1985 to 1995, the rate of access to the general baccalauréat has levelled out to some extent since then, whereas the rate rose for the technology and vocational baccalauréats and then stabilized. These latter baccalauréats, especially the vocational ones, were in practice designed for the least academic pupils: given the way the career choice system works in France, all pupils who are capable of doing so choose to stay as long as possible in the general track. Since more working-class pupils have academic difficulties, they are over-represented on the technology and vocational tracks: 14% of students following the general track are children of manual workers, compared with 26% in the technology track and 36% in the vocational track. Children with unemployed or inactive parents represent 5%, 8% and 12% respectively of students in these various tracks.

Table 1.

Percentage of age group achieving various types of baccalauréat

	General baccalauréat	Technology baccalauréat	Vocational baccalauréat	All
Note: percentages are obtained by dividing the number of baccalauréat holders by the total population of the age group in each year.
Source: Ministry of Education (Repères et Références Statistiques, 2006 edition).
1970	16.7	3.4		20.1
1985	19.8	9.6		29.4
1990	27.9	12.8	2.8	43.5
1995	37.2	17.6	7.9	62.7
2000	32.9	18.5	11.4	62.8
2005	33.7	17.3	11.5	62.5

It is therefore justifiable to speak of "segregative democratization" (Merle, 2000 and 2002), since greater access to the baccalauréat has involved differential development of access to the various types. These differences may be seen as inequalities, from the perspective adopted here, in the sense that the rates of entry into and success in higher education correlate strongly with the type of baccalauréat obtained, from the general, with by far the best results, down to the vocational.

Development of inequalities in access to the baccalauréat in recent years

To assess these developments, we took two birth cohorts from the FQP survey, 1962-1967 (who sat the baccalauréat in the early 1970s) and 1975-1980 (baccalauréat in around 1995); these two groups were selected as "end markers" of the period when the proportion of baccalauréat-holders and the access rate to higher education rose fastest⁽⁶⁾. We distinguish between the probability of passing a baccalauréat (of whatever type) and the probability of passing a particular type of baccalauréat (general, technology or vocational).

To measure the extent of the inequalities, we use the two types of indicator common in the literature: differences between rates (additive model) and odds ratios. Without going into the debate on the relevance and meaning of the two measurements (for detailed discussion, see Duru-Bellat and Kieffer, 2001; Manzo, 2006), we merely recall that additive models are relevant for the view that any extension of access to an educational good is to be seen as democratization, because education is a good in itself; one is therefore justified in examining how much education each person receives. The odds ratios express a different vision, whereby education is a positional good, whose value is only relative to what others have. It is the relative steepness of changing access rates for various groups that is important, revealing developments in competition between those groups. For the educational level we are considering, the baccalauréat, both perspectives are legitimate⁽⁷⁾: the baccalauréat has a highly symbolic value in itself in France and betokens knowledge of intrinsic value; but in terms of the career openings it provides, a relativistic approach is more appropriate.

Table 2 presents the changes in rates of access to the baccalauréat observed for the two birth cohorts, using the extreme cases of the children of persons in higher-level occupations (6.4% of the total in the first case and 14.6% in the second) and the children of manual workers (39.3% and 36.4%).

Having a baccalauréat or not

The odds ratios first demonstrate a certain degree of democratization in access to the baccalauréat, when no distinction is made between types of qualification, in which case the ratio falls from 11.3 to 6.3 (between children of parents in higher-level occupations and manual workers' children). This development is also visible (indeed much more striking) in the percentage point differences, since the gap between the two groups falls from 54 to 33: although competition decreases only slightly, the increase

Table 2.

Access of manual workers' children to various types of baccalauréat compared with children of persons in higher-level occupations

	Percentage with baccalauréat			Difference between children of persons in higher-level occupations and manual workers' children
	All	Children of persons in higher-level occupations	Manual workers' children	Percentage point difference	Odds ratio
Interpretation: in the 1962-1967 birth cohort, children of parents in higher-level occupations are 11.3 times more likely than the children of manual workers to obtain a baccalauréat than not.
Source: authors' calculations from FQP 2003 survey.
1962-1967 birth cohort
All baccalauréat-holders	35.0	74.5	20.6	53.9	11.3
General baccalauréat	22.0	60.9	9.8	51.1	14.3
Technology or vocational  baccalauréat	13.0	13.6	10.8	2.8	1.3
1975-1980 birth cohort
All baccalauréat-holders	63.4	89.2	56.6	32.6	6.3
General baccalauréat	38.6	74.0	20.2	53.8	11.2
Technology or vocational baccalauréat	24.9	15.2	36.4	–21.2	0.3

in the proportion of baccalauréat-holders makes the acquisition of this "good" less inegalitarian.

Logistic regressions, permitting analysis of variables with "all other things being equal", confirm this first observation (Table 3). Between the two birth cohorts, the coefficients of the impact of father's occupation on the chances of holding a baccalauréat fall (although they are still significant at the end of the period); in particular, the advantage for children of parents in higher-level occupations decreases, as is also the case, to a lesser extent, for the children of people in intermediate occupations and of farmers. Again taking manual workers' children as the reference, we note that the children of clerical workers noticeably approach that category in all the models. In terms of the influence of parents' educational level, having a father with a degree in higher education provides a major and growing advantage. The gap between the children of unqualified fathers and of fathers with the CAP or BEP vocational certificates is slightly reduced, no doubt because of the declining relative status of the CAP in the educational hierarchy of the parents' generation. The weight of unqualified fathers fell over the period (from 60% to 42%), thus accentuating the singular characteristics of this group.

The influence of the mother's occupation appears to be greater in the second cohort (particularly the advantage of having a mother in a higher-level occupation), no doubt because of the increase in women's labour force participation. But the impact of the mother's educational level does not increase (or is perhaps increasingly "absorbed" by the increasing effect of her occupation). At all events, these results are in line with some earlier observations suggesting that the structure of social inequality may have changed in recent years, with a decreasing influence of the father's occupation and an increasing influence of his educational level (see, for example, Goux and Martin, 1997). Inequalities would appear to be changing shape rather than disappearing, and this is no doubt also true of the mechanisms behind them.

This development is a recent one. Comparable models estimated for cohorts born from the 1930s onwards displayed a general stability of inequalities in access to the baccalauréat until those born in the late 1950s (Duru-Bellat and Kieffer, 2001). By contrast, comparing the two cohort groups in this study, major changes can be seen in parallel with the spectacular increase in the rate of access to the baccalauréat. But the downward trend in the influence of socio-occupational origin is combined with an increasing role of parents' educational level, more marked as fewer people fail to reach baccalauréat level.

The greater access to the baccalauréat does not reduce girls' advantage, which indeed has greatly increased over the period⁽⁸⁾. As the regressions show, girls have especially benefited from the diversification of baccalauréats, since their advantage has decreased only for the general baccalauréat while increasing overall (all types).

What sort of baccalauréat?

We turn now to the type of baccalauréat obtained, since this is an important factor in determining the rate of entry into higher education, the choice of courses available and later success rates. Compared with the other types of baccalauréat, the general baccalauréat remains more inegalitarian⁽⁹⁾ (Table 2), but inequality is falling, although it remains at a high level. Between the two cohorts, the odds ratio falls from 14.3 to 11.2: there is a degree of democratization in achieving a general baccalauréat, even if the percentage point differences increase slightly. This is due to the fact that the access rate for children of parents in higher-level occupations, even if it rises more in percentage points than that of manual workers' children, cannot double because it is already

Table 3.

Access to the baccalauréat and the general baccalauréat across the entire cohort, and to the general baccalauréat among baccalauréat-holders (logit model coefficients)

	Model 1 All types of baccalauréat (whole population)		Model 2 General baccalauréat (whole population)		Model 3 General baccalauréat among baccalauréat-holders
	1962-1967 cohort	1975-1980 cohort	1962-1967 cohort	1975-1980 cohort	1962-1967 cohort	1975-1980 cohort
Significance levels: *: p ≤ 0.05; **: p ≤ 0.01; ***: p ≤ 0.001.
Interpretation: the estimated coefficient for individuals defined by an active modality indicates the influence of that modality, all other things being equal, compared with the individuals defined by the reference modality; G2 (log likelihood) estimates the explanatory power of the model.
Source: authors' calculations from FQP 2003 survey.
Father's occupation
Manual worker (ref.)	0	0	0	0	0	0
Farmer	0.56 ***	–0.10	0.35	0.32	–0.14	0.42
Self-employed: business, trade and crafts	0.72 ***	0.67 ***	0.74 ***	0.95 ***	0.24	0.82 ***
Higher-level occupation	1.44 ***	0.82 ***	1.46 ***	1.10 ***	0.71 ***	0.94 ***
Intermediate occupation	0.96 ***	0.75 ***	1.03 ***	0.68 ***	0.49 ***	0.44 **
Clerical or sales worker	0.45 ***	0.22 *	0.71 ***	0.34 ***	0.63 ***	0.28 *
Other	0.11	–0.35 *	0.54 **	–0.64 ***	0.90 **	–0.62 *
Mother's occupation
Economically inactive (ref.)	0	0	0	0	0	0
Farmer	0.10	1.10 **	0.09	0.32	0.03	–0.29
Self-employed: business, trade and crafts	0.10	0.09	0.08	–0.30	0.11	–0.48 *
Higher-level occupation	–0.03	0.99 **	0.005	0.41	0.16	0.06
Intermediate occupation	0.17	0.49 **	0.17	0.41 **	0.09	0.26
Clerical or sales worker	–0.03	0.31 ***	–0.04	0.23 *	0.01	0.10
Manual worker	– 0.07	0.07	–0.43 *	–0.12	–0.56 *	–0.19
Father's qualification
CAP, BEP, BEPC (ref.)	0	0	0	0	0	0
None or primary certificate	– 0.47 ***	– 0.27 **	– 0.28 ***	– 0.28 **	0.15	– 0.14
Baccalauréat	0.30 *	0.30 *	0.47 **	0.37 **	0.51 **	0.34 *
Baccalauréat + 2 years	0.27	0.94 ***	0.67 ***	0.77 ***	0.93 **	0.54 **
Degree in higher education	0.51 **	1.13 ***	1.05 ***	1.00 ***	1.48 ***	0.77 ***
Mother's qualification
CAP, BEP, BEPC (ref.)	0	0	0	0	0	0
None or primary certificate	–0.49 ***	–0.56 ***	–0.48 ***	–0.49 ***	–0.12	–0.26 *
Baccalauréat	0.57 ***	0.52 ***	0.44 ***	0.42 ***	0.15	0.25
Baccalauréat + 2 years	0.46 *	0.42 *	0.49 ***	0.72 ***	0.41	0.77 ***
Degree in higher education	0.77 **	0.24	0.74 ***	0.68 ***	0.28	0.84 **
Sex
Male (ref.)	0	0	0	0	0	0
Female	0.35 ***	0.58 ***	0.66 ***	0.67	0.71 ***	0.51 ***
Constant	–0.78 ***	–0.06	–1.98 ***	–1.48 ***	–0.54 **	–0.52 ***
G2	6,706	4,309	5,161	4,113	2,312	2,887
Sample size	5,984	3,889	5,984	3,889	2,111	2,454

[Begin Page 128]

high. Furthermore, the greater baccalauréat access rate among children of parents in higher-level occupations affects only the general baccalauréat courses, whereas among manual workers' children the increase is spread across all types of baccalauréat, particularly the vocational baccalauréat.

Logistic regressions (Table 3) comparing the cohorts confirm this change. The influence of the father's occupation on the probability of passing a general baccalauréat appears to decline, whereas for all the other parameters (mother's occupation, both parents' educational levels, sex), the values are mainly stable.

For the technology or vocational baccalauréat, matters are quite different. Since these examinations were created, they have always appealed more to working-class students than the general baccalauréat; the odds ratios and even more the percentage point differences (Table 2) show that they appeal increasingly to the working class as the baccalauréat access rate increases (in the second cohort group). This may be compared with the previous observation: as the general baccalauréat becomes more common (along with access to the lycée), those students who still move on to vocational (BEP) courses at the end of year 9 are increasingly working-class in origin, since the social inequalities in academic achievement persist. Here the appropriate term is segregative democratization.

Among baccalauréat holders, holding one type rather than another

The previous observations, based on the whole of each cohort group, combine two phenomena: the opening up of secondary education on the one hand, and changes, if any, in career orientation mechanisms on the other. To separate these out, it is useful to specifically examine the category of baccalauréat-holders. As Table 4 shows, manual workers' children at this level are more likely to have a vocational baccalauréat in the second cohort; they are also less likely to have a general or technology baccalauréat, whereas the distribution of children of parents in higher-level occupations among types of baccalauréat has hardly altered. The odds ratios clearly demonstrate the change: access to the general baccalauréat increases (from 4.7 to 6.2), along with the percentage point differences (from 33 to 39). There is indeed an accentuation of social inequality in access to the general baccalauréat among baccalauréat-holders. However, when the probability of access to a technology baccalauréat is examined separately, there is a very slight reduction in social inequality (percentage point differences from – 30 to – 21). It is consequently the development of the vocational baccalauréat – which offers a chance of entering a baccalauréat course after a short vocational course (CAP-BEP) – that explains the increase in inequality of access to the general baccalauréat, since manual workers' children constitute most of the course entrants.

Table 4.

Distribution of baccalauréat-holders by type of baccalauréat (%)

	Percentage of baccalauréat-holders			Difference between children of parents in higher-level occupations and manual workers' children
	All	Children of parents in higher-level occupations	Manual workers' children	Percentage point difference	Odds ratio
Source: authors' calculations from FQP 2003 survey.
1962-1967 birth cohort
General baccalauréat	63	82	49	33	4.7
Technology baccalauréat	30	13	43	–30	0.2
Vocational baccalauréat	7	5	8	–3	0.6
1975-1980 birth cohort
General baccalauréat	62	83	44	39	6.2
Technology baccalauréat	23	11	32	–21	0.3
Vocational baccalauréat	15	6	24	–18	0.2

The model explaining the probability of obtaining a general baccalauréat among baccalauréat-holders alone (Table 3, Model 3) confirms that access to this qualification has not been democratized; on the contrary, the influence of the father's occupation has increased, slightly for higher-level occupations and more so (upwards) for the self-employed, while clerical and sales workers tend to come closer to manual workers; the positive influence of the mother's educational level has also increased, particularly at the higher levels, as access to the baccalauréat became so general as to include a majority of the age group: passing the general baccalauréat or not is increasingly correlated with the mother's educational level, even though the coefficients of the influence of the father's qualification are falling.

The use of the term segregative democratization is therefore justified: democratization occurred mainly via the increase in vocational baccalauréats, while at the same time it became relatively harder for children from less educated backgrounds to acquire a general baccalauréat.

Within a few decades, therefore, the type of baccalauréat obtained has become a major vector of social inequality: all pupils stay longer in the school system but they are directed towards a clear hierarchy of tracks. Once a large proportion of an age group passes the baccalauréat, these pupils will be differentiated by more specific criteria. Examination of social inequality must therefore focus not only on the type of baccalauréat but also the subject choice and grade obtained.

Passing a science baccalauréat, gaining a good grade

For our purposes it is useful to examine in particular the science baccalauréat, which at present enjoys the highest prestige and offers the most career opportunities(10): the odds ratio falls from 15.9 to 6.4 between cohorts. Social inequality with respect to this option is noticeably lower than for the general baccalauréat as a whole and the democratization process more marked, whereas one might have supposed that this prestige option would be relatively protected against the general opening-up. This equalizing democratization is confirmed by Model 1 in Table 5, where the coefficients expressing the influence of the father's socio-occupational category are lower.

Among baccalauréat-holders, the proportion of science baccalauréats is stable at 30.6% of the first cohort of baccalauréat-holders and 29.4% of the second. At a time when the numbers taking the examination have increased, the relative position of the science option has not increased. However, there is a certain degree of democratization: 48.4% of children of parents in higher-level occupations obtained a science baccalauréat in the first cohort, compared with 16.7% of manual workers' children. In the second cohort, the figures are 44.2% and 19.7%. The odds ratios reflect this change, falling from 4.7 to 3.2. Model 2 in Table 5 shows a reduction in the influence of the father's occupation and educational level (neither characteristic of the mother is significant in either cohort).

If we examine only holders of the general baccalauréat, the figures again show a clear narrowing of the gap between social groups. In the first cohort, 59% of children of parents in higher-level occupations obtain a science baccalauréat compared with 35% of manual workers' children. In the second cohort, these figures are 53% for the former and 45% for the latter. The odds ratios confirm this observed reduction in inequality, falling from 2.7 to 1.2. Since the models explain the acquisition of a science rather than non-science baccalauréat among only those who hold a general baccalauréat (Table 5, Model 3), they show that at this level social origin no longer has a significant influence; the fact of having a father with a degree in higher education was significant in the first cohort but not in the second. Children of working-class origin (especially from farming backgrounds, the best at least, who are not directed towards the vocational baccalauréat) take advantage of access to the general baccalauréat to opt for a science baccalauréat, either because they are particularly talented (having been relatively over-selected to study for a general baccalauréat) or because the content or career opportunities of the sciencebaccalauréat seem more attractive(11). Similarly, the new baccalauréat-holders who are children of parents in higher-level occupations, probably weaker academically (since their predecessors did not pass the baccalauréat, in a social group where this was the rule) have chosen options other than the science one. From the standpoint of the degree of academic selection represented by access to a general baccalauréat, it would be wrong to speak of segregative democratization; this term should be reserved for what happens in access to particular types of baccalauréat. On the contrary, this situation appears to be tending towards meritocracy.

This observation would need to be qualified by considering the subject choices available within each baccalauréat option, and specifically the science option, but the FQP survey data cannot be used for that purpose. However, data from the evaluation, perspective and performance department (DEPP) of the Ministry of Education show, for example, that based on the panel of pupils who entered year 6 in 1995, 28.5% of children of parents in higher-level occupations obtained a science baccalauréat with maths and physics, while only 4.6% of manual workers' children did so (odds ratio: 8.2).

In the science baccalauréat, girls and boys differ in that girls less often take this option. This is true when one examines the whole cohort (to a slight extent, since more girls take the baccalauréat) or all baccalauréat-holders, and even more so for holders of the general baccalauréat alone (this is the only strongly significant factor among general baccalauréat-holders). At all three levels of analysis, the phenomenon tends to increase from one cohort to the other.

Another indicator of academic excellence is relevant here: a good grade, namely an average mark of at least 12 out of 20 in the baccalauréat examinations. This is particularly important for certain selective courses in higher education. The likelihood of gaining a good grade is higher with general baccalauréats (29.5% and 35.3% in the two cohorts) than with technology baccalauréats (13% and 25.7%), but it is now particularly high with vocational baccalauréats (19.6% and 41.8%). The top grades (bien and très bien, for grade averages above 14 and 16 respectively), which lead to the most prestigious "elite" tracks, are given more often in the general baccalauréat (13.6% of the second cohort, compared with less than 5% in the technology baccalauréat and 9.8% in the vocational baccalauréat). Good grades are more frequently achieved by children of parents in higher-level occupations, confirming early observations by Convert and Pinet (1989); among the general baccalauréat-holders in the second cohort, 41.3% of children of parents in higher-level occupations obtained a good grade, compared with 29.4% of manual workers' children. Nevertheless, at this stage of selection, social inequality is becoming much less marked, as Table 6 shows.

Table 5.

Influence of individual characteristics on the chances of passing a science baccalauréat (logit model coefficients)

	Model 1 Whole cohort		Model 2 Baccalauréat-holders		Model 3 General baccalauréat-holders
	1962-1967 cohort	1975-1980 cohort	1962-1967 cohort	1975-1980 cohort	1962-1967 cohort	1975-1980 cohort
Significance levels: *: p ≤ 0.05; **: p ≤ 0.01; ***: p ≤ 0.001.
Interpretation: The estimated coefficient for individuals defined by the active modality indicates the influence of that modality, all other things being equal, compared with the individuals defined by the reference modality; G2 (log likelihood) estimates the explanatory power of the model.
Source: authors' calculations from FQP 2003 survey.
Father's occupation
Manual worker (ref.)	0	0	0	0	0	0
Farmer	0.56 *	0.80 **	0.112	0.94 **	0.45	0.97 *
Self-employed: business, trade and crafts	0.93 ***	0.70 ***	0.40 *	0.43 *	0.34	–0.12
Higher-level occupation	1.36 ***	0.77 ***	0.59 **	0.50 **	0.31	–0.006
Intermediate occupation	–1.03 ***	0.48 **	0.41 *	0.20	0.16	–0.18
Clerical or sales worker	0.84 ***	0.15	0.58 **	–0.002	0.24–	0.33
Other	0.56	–0.48	0.40	–0.37	–0.05	0.05
Mother's occupation
Economically inactive (ref.)	0	0	0	0	0	0
Farmer	0.42	–0.36	0.53	–0.88	0.81	–0.95
Self-employed: business, trade and crafts	0.11	–0.32	0.11	–0.40	0.10	–0.20
Higher-level occupation	0.16	0.32	0.20	0.19	0.15	0.25
Intermediate occupation	0.19	0.33 *	0.12	0.25	0.11	0.20
Clerical or sales worker	–0.09	0.09	–0.05	0.003	–0.09	–0.11
Manual worker	–0.35	–0.17	–0.23	–0.19	0.07	–0.19
Father's qualification
CAP, BEP, BEPC (ref.)	0	0	0	0	0	0
No qualification or CEP	–0.30 *	–0.43 **	–0.03	–0.34	–0.18	– 0.15 **
Baccalauréat	0.54 **	0.18	0.48 *	0.09	0.28	–0.17
Baccalauréat + 2 years	0.80 ***	0.54 **	0.85 ***	0.33 *	0.56 *	0.03
Degree in higher education	1.08 ***	0.70 ***	1.06 ***	0.50 **	0.56 *	0.17
Mother's qualification
CAP, BEP, BEPC (ref.)	0	0	0	0	0	0
No qualification or CEP	–0.38 **	–0.52 ***	–0.10	–0.29*	0.02	–0.12
Baccalauréat	0.29	0.28 *	0.07	0.15	–0.003	0.02
Baccalauréat + 2 years	0.48 *	0.36 *	0.34	0.28	0.21	–0.03
Degree in higher education	0.73 **	0.10	0.61 *	0.01	0.55	–0.44
Sex
Male (ref.)	0	0	0	0	0	0
Female	–0.16	–0.41 ***	–0.44 ***	–0.65 ***	–1.12 ***	–1.29 ***
Constant	–2.69 ***	–1.68 ***	–1.30 ***	–0.82 ***	0.20	0.95 ***
G2	3,348	3,182	2,349	2,692	1,646	1,860
Sample size	5,984	3,889	2,111	2,454	1,306	1,465

[Begin Page 133]

Table 6.

Inequality between children of parents in higher-level occupations and manual workers' children at various academic levels (odds ratios)

	Percentage of study group obtaining the baccalauréat	Percentage of baccalauréat- holders obtaining a general baccalauréat	Percentage of general baccalauréat- holders obtaining a science baccalauréat	Percentage of science baccalauréat- holders obtaining a good grade
Source: authors' calculations from FQP 2003 survey.
1962-1967
cohort	11.4	4.7	2.7	1.3
1975-1980
cohort	9.6	6.2	1.2	1.4

The group that is ultimately eligible for entry to higher education is thus gradually stratified according to the type and option of baccalauréat and the grade obtained. This segmentation is both academic and social. Table 6 clearly shows that a large part of the social selection occurs before access to the baccalauréat as a combined result of unequal achievement and unequal career orientation. Among those who obtain the baccalauréat, there is already slightly less inequality in the chances of obtaining a general baccalauréat. Among those who obtain the general baccalauréat, it is as if social inequality was this time outweighed by actual academic inequality: unequal degree of achievement reflected in the choice of a science option, and then a good grade in the science option.

What will be the effect of these developments in inequality of access and career orientation for higher education? Once the baccalauréat becomes less selective, one may expect (as many authors have maintained; for example, Mare, 1981) that a wider and more diverse student group will reach this higher stage of career choice and that the social inequality that faded at the lower stage will now increase, either in terms of overall access rate or in the trajectories followed in higher education.

II. Changing transitions from secondary to higher education

Since the baccalauréat is a passport to higher education, the final stage of the educational process, the issue of democratization now shifts to this final level. First, we describe the changes observed between the two cohort groups, with respect to whole of the study group, i.e. taking into account the opening-up of the secondary system and the democratization of educational careers that has resulted at earlier levels. Then we focus on developments among baccalauréat-holders.

Chances of entering higher education and obtaining a degree

Let us first examine the rates of access to higher education for the whole of the study group. These rates increased sharply in the short period between the two cohorts (from 27.7% of an age group to 53.2%). However, social inequality changed little.

Since all social groups benefited in a comparable manner from this expansion, percentage point differences and odds ratios (Table 7) do decline but only slightly (11.4 to 11.2). We are in the situation that P. Merle (2000) qualifies as "uniform democratization", where higher access rates are associated with stable inequalities. The significant widening of access to higher education appears not to have reduced social inequalities, probably because the rates have not yet reached a ceiling. This is likely to happen soon, however, since the rate of access of children of parents in higher-level occupations already exceeds 85% in the second cohort.

This stability is relatively surprising since there has been a noticeable democratization in achieving the baccalauréat among the general population

Table 7.

Inequality of access to higher education

	Percentage access			Difference between children of parents in higher-level occupations and manual workers' children
	All	Children of parents in higher- level occupations	Manual workers' children	Percentage point difference	Odds ratio
Source: authors' calculations from FQP 2003 survey.
Whole cohort
1962-1967 cohort	27.7	66.1	14.6	51.5	11.4
1975-1980 cohort	53.2	85.3	34.2	51.1	11.2
Baccalauréat-holders
1962-1967 cohort	75.3	85.1	67.7	17.4	2.7
1975-1980 cohort	81.7	92.9	70.7	22.2	5.4
General baccalauréat-holders
1962-1967 cohort	88.6	92.6	84.3	8.3	2.3
1975-1980 cohort	95.4	95.9	93.0	2.9	1.8
Technology and vocational baccalauréat-holders
1962-1967 cohort	52.8	51.6	52.6	–1.0	1.0
1975-1980 cohort	60.4	77.8	53.6	24.2	3.0

and very many baccalauréat-holders go on to higher education. But the population of baccalauréat-holders is highly diverse. Rates of access to higher education are indeed close to 100% for pupils with a general baccalauréat: in the first cohort group, 86.6% went on to higher education, as did 95.4% from the second cohort. The figures are lower for holders of a technology baccalauréat, but did increase from 56% in the first cohort to 78% in the second. In the case of vocational baccalauréat-holders, figures were low at the start of the period, since the qualification was created in 1987, so a reliable estimate of the change between the two cohorts cannot be obtained (access rate rising from 3.5% to 6%)⁽¹²⁾.

To understand why democratization in obtaining the baccalauréat is occurring alongside a practically unchanging situation of unequal access to higher education, it is necessary to consider the diversity of baccalauréat-holders and specifically examine the rates of access and choices made by various types of student.

Taking baccalauréat-holders as a whole, the rate of access to higher education among children of parents in higher-level occupations has risen (nearly 93% in the second cohort) while the increase remains low for manual workers' children. The overall improvement in access to higher education for baccalauréat-holders has not closed the gap, quite the contrary: the percentage point differences have risen and odds ratios even more. The odds ratio is 5.4 at the end of the period (compared with 11.2 for the entire population)⁽¹³⁾. Within the group of baccalauréat-holders alone (all types of baccalauréat), there is an increase in the social inequality of access to higher education. It is as if the disappearing inequalities in access to the baccalauréat re-emerge after that point; in other words, holding a baccalauréat now guarantees less than before a socially homogenous entry into higher education, because of diversification of the diploma itself, with the development of the technology and vocational baccalauréats.

Taking general baccalauréat-holders alone, there is a slight reduction in inequality (although almost all go on to higher education), which is relatively small at this stage. The previous observation is therefore not explained by more marked social inequalities within each type of baccalauréat, but rather by developments in inequality that affect the choice of baccalauréat, a point to which we return below.

Taking technology and vocational baccalauréat-holders together, there is an increase in social inequality due to the development of the vocational baccalauréat, which is less suited to entry into higher education than the technology baccalauréat, and also more attractive to pupils with working-class origins.

What then are the chances of gaining a university degree, which is essential for job-seeking purposes? Here the odds ratio between children of parents in higher-level occupations and manual workers' children falls between the two birth cohorts (1962-1967 and 1972-1977⁽¹⁴⁾), from 10.9 to 9.2. This is in line with the trend of democratization noted for the earlier diploma, the baccalauréat. It is as if the "wave of democratization" in France that began earlier with greater access to collège, then lycée (Duru-Bellat and Kieffer, 2001) has now reached higher education: the chances of gaining a degree in higher education have become less unequal between social groups since the 1980s, and examination of more recent birth cohorts (1975-1980) shows that the trend appears to be continuing. In the case of our study, there were still wide differences in rates of access at the end of the study period, since only 17.2% of manual workers' children in the second cohort gained a degree in higher education compared with 65.6% of children of parents in higher-level occupations.

Next, specific examination of the group of baccalauréat-holders alone reveals developments in access both to the baccalauréat and to careers in higher education (in other words gaining entry and obtaining a degree). From this standpoint (degrees gained by baccalauréat-holders alone), the odds ratios are three times smaller, because most of the social inequality had its effect earlier, when the student passed the baccalauréat or not. In particular, the odds ratio rises slightly between the cohorts (from 2.8 to 3.3), a slight increase explained by the earlier observation, namely the increase in inequality among baccalauréat-holders going on to higher education.

An initial conclusion is that the democratization in the acquisition of a degree in higher education, which is real, is entirely due to the expansion of the qualification that provides entry into higher education, the baccalauréat; access to higher education in itself has not been democratized, far from it, as there has been greater inequality in passing various types of baccalauréat. In other words, this democratization has been segregative.

Finally, let us examine inequality in the most selected group, general baccalauréat-holders. In absolute terms, inequality here is lower and tending to decrease over time; both percentage point differences and odds ratios fall slightly, with an odds ratio of 2.1 at the end of the period; it cannot therefore be said that inequality in the chances of gaining a degree has significantly changed for this sub-group.

Stages in the selective process up to a degree in higher education

All these results highlight the changes in the mechanisms of social inequality that have accompanied the growth in access to education over the last twenty years. There has been a major expansion in each age group's access to the baccalauréat, but this has not led to a proportional democratization of access to higher education, because manual workers' children have only seen their chances of obtaining a baccalauréat improve at the cost of being relegated to the technology and vocational options. This has increased inequality within the group of baccalauréat-holders for two reasons: first because these technology and vocational baccalauréat-holders less often go on to higher education, and second, because their chances of succeeding if they do are much lower, given that they are selected out by (academic) failure⁽¹⁵⁾. On the other hand, those students from working-class backgrounds who passed general baccalauréats (a smaller proportion than for other baccalauréats) saw their chances of going on to higher education come closer (slightly) to those of children of parents in higher-level occupations.

Similarly, it is clear that although overall the acquisition of a degree in higher education has been democratized, this democratization is entirely due to that of the baccalauréat. If we implement a conditional approach, i.e., if we focus on baccalauréat-holders alone, there is no democratization; indeed the trend is rather a slight increase in inequality. This slight increase disappears if general baccalauréat-holders are taken separately. Consequently, the only mechanism that explains the rise in the inequality of chances of obtaining a degree in higher education is indeed the choice of type of baccalauréat (with the limitations this may place on subsequent educational opportunities). As access to education has widened, working-class children do actually pass the baccalauréat more often than before, but they are directed more firmly towards the technology and even more the vocational options, which narrows the opportunities for going on to higher education and gaining a degree. On the other hand, when working-class children gain access to a general baccalauréat, social inequality again tends to fall slightly, and the models show that it is no longer significant.

Our results differ partly from those of Selz and Vallet (2006), based on employment surveys, who sought to explain long-term inequality in relation to the highest level of qualification obtained. They first observe that for all cohorts and those born between 1920 and 1976 there has been a reduction in social inequality, concentrated on the 1938-1946 cohorts, but less marked for subsequent cohorts. Admittedly, our results cover a shorter, more recent period and only concern the acquisition of a degree in higher education, but the general trend towards democratization is confirmed. When these authors separate out baccalauréat-holders, their observation is reversed, and inequality increases, which agrees with our results. However, for the period they study, they obtain rising inequality for general baccalauréat-holders alone, whereas we conclude that there is a slight decrease in inequality. This causes us to interpret these changes as due mainly to the shifts that have taken place between the various types of baccalauréat. Selz and Vallet, on the other hand, attribute them more to the increasing diversity of baccalauréat-holders, a complementary rather than competing interpretation⁽¹⁶⁾.

Whereas social selection until the mid-1980s (before the massive expansion in the number of baccalauréat candidates) involved finishing secondary education and obtaining a qualification at the end of a school career that might end rapidly, after the choices made in year 7 and year 9, it is now more a matter of the type of baccalauréat one studies for (general, technology or vocational) even if, clearly, earlier education and the associated inequality continue to have an influence, since they affect the choice made between these options.

The hypothesis of Arum et al., (2007) mentioned above, is apparently confirmed by these initial results: because the various types of baccalauréat are differentiated not "horizontally" but "vertically", working-class children are diverted from the opportunities offered by the general baccalauréat, a fact which, since secondary and higher education are closely linked, has a major impact on their later educational careers. This diversion is reminiscent of remarks made by Cherkaoui (1982), after the technology baccalauréats were created in the late 1960s, who observed that the new baccalauréats in practice channelled working-class children towards these options, thereby reducing their chances of studying for the general baccalauréats that had existed until then.

But the type of baccalauréat obtained and the constraints this places on further education are not the only form of differentiation between educational trajectories. It is useful, albeit rarely found in the literature, to extend the enquiry into the types of study and specialities within higher education, as can be done with the FQP survey data.

III. Choices made in higher education and the associated social inequality: is differentiation increasingly horizontal?

To analyse development in the choices made in higher education, it is legitimate to examine the first choice, even if it is known that not only will there be other choices in the case of failure, but that some students adopt deliberate strategies to maximise their chances of achieving a given "final" degree, for example by entering an IUT (Institut Universitaire de Technologie) with a view to joining an engineering school or university course after graduation. Theoretically, students in France may choose any course with any type of baccalauréat. But within this general principle, there are explicit selection procedures for particular tracks – CPGE, IUT, STS courses in particular⁽¹⁷⁾ – and a process of self-selection by students, well attested in the literature (Duru Bellat, 2002). Possession of a science baccalauréat maximizes the chances of being admitted to a selective course, even though there is some self-selection, including among the academically most successful students, particularly on the basis of their background or sex. The following analyses take account of these parameters.

Access to various types of study: the first "major choice"?

Higher education in France cannot be considered as a single undifferentiated whole. There are quite distinct types of course with unequal levels of student selection and involving varying types of vocational study. We first examine the division between selective and non-selective courses, placing together as "elite" selective courses the CPGEs, medical faculties, and engineering schools which directly admit baccalauréat-holders; these we distinguish from all the other non-selective university courses (except medicine) and short selective vocational courses (IUT, STS, paramedical and social courses, etc.).

When all cohorts are examined for the overall chances of access to a particular higher education diploma, democratization can be seen at work in all types of diploma, since the odds ratios all decline from one to the other (Table 8). Note that the same result is found when we focus not on access but on acquisition of final diploma. We see the extent to which access to elite higher education in France remains deeply inegalitarian (with an odds ratio of 12.1 in the second cohort), whereas democratization is advancing for shorter courses.

On the other hand, taking baccalauréat-holders alone, the conclusion is quite different: democratization is hardly the right term to use. Admittedly, short vocational higher education courses are attracting more working-class

Table 8.

Access of manual workers' children to various types of higher education compared with children of parents in higher-level occupations

	Percentage access			Difference between children of parents in higher-level occupations and manual workers' children
	All	Children of parents in higher-level occupations	Manual workers' children	Percentage point difference	Odds ratio
Source: Author's calculations based on 2003 FQP survey.
Whole cohort
1962-1967 cohort
Vocational higher education	13.4	23.5	7.8	15.7	3.6
University except medicine	9.9	23.3	5.5	17.8	5.2
Elite higher education	4.4	19.3	1.3	18.0	18.2
1975-1980 cohort
Vocational higher education	25.3	30.3	19.5	10.8	1.8
University except medicine	21.1	34.4	12.6	21.8	3.6
Elite higher education	6.9	20.6	2.1	18.5	12.1
Baccalauréat-holders
1962-1967 cohort
Vocational higher education	35.8	16.6	23.0	–6.4	0.7
University except medicine	27.3	22.3	21.6	0.7	1.0
Elite higher education	12.1	41.8	12.3	29.5	5.1
1975-1980 cohort
Vocational higher education	38.3	17.2	27.9	–10.7	0.5
University except medicine	32.6	23.9	21.5	2.4	1.1
Elite higher education	10.7	43.2	11.1	32.1	6.1
General baccalauréat-holders
1962-1967 cohort
Vocational higher education	34.0	21.5	18.4	3.1	1.2
University except medicine	36.5	24.6	18.3	6.3	1.5
Elite higher education	18.0	44.0	10.1	33.9	7.0
1975-1980 cohort
Vocational higher education	33.2	23.8	21.4	2.4	1.1
University except medicine	46.1	25.1	19.4	5.7	1.4
Elite higher education	16.1	45.9	10.7	35.2	7.1

students (which is clearly related to the fact that working-class children are increasingly likely to obtain a technology or vocational baccalauréat), but elite higher education is actually becoming more socially selective, whereas the situation in the universities is stable and moving close to equality. Here too, the term segregational democratization seems appropriate.

This observation is probably related to the unchanging weight of the elite sector. The relative broadening of access to the various sectors, like the relative opening up at secondary level (which determines the number of students eligible for higher education), are parameters that affect changes in social inequality. Recent studies (Albouy and Wanecq, 2003) have shown that the democratization of higher education in general has, as it were, spared the elite grandes écoles which, in a context of expanding student numbers, have managed to maintain a constant relative weight. There has even been an accentuation of social inequality in access to the most prestigious grandes écoles.

Finally, taking general baccalauréat-holders alone, a sharp contrast can be seen between sectors where equality is virtually achieved (short vocational courses and, to a lesser extent, non-selective university courses) and the elite sectors, which remain highly inegalitarian (odds ratio: 7.1). The selected group of general baccalauréat-holders go on to higher education almost in proportion to their social origins, but there is still inequality between sectors, at least for the elite sector, which resists change. Although odds ratios here vary little between cohorts, they tend downwards for short vocational and university courses and upwards in elite higher education.

In this unequal access to elite higher education, it is interesting to determine the specific role of the educational capital accumulated by children of parents in higher-level occupations with respect to the role of course choices that appear to differ significantly between students of comparable academic achievement. To shed light on this question, we estimated logit models explaining students' initial choices (which may differ from diplomas finally achieved) for the various courses on offer. To the extent that one may consider the courses to be ranked (in academic hierarchy and employment opportunities), we first distinguished the elite sector from all other courses, and then, taking only those students who did not get into the elite sector, we separated short vocational courses from university ones (Table 9)⁽¹⁸⁾.

The first comparison depends primarily on the type of baccalauréat obtained, since science baccalauréats are by far the most dominant. For comparable baccalauréats, the father's level of qualification (but not his occupation) also has a major influence: the student's choice of the elite sector is greater the more educated the father. Fewer girls choose this sector. From one cohort to the other, the influence of the father's educational level tends rather to increase and a positive impact of being a farmer's child also emerges (no doubt due to sociological changes in this group over the study period). The second comparison (among those who failed to or did not want to enter the elite sector) does not

Table 9.

Course choice according to student characteristics (logit model coefficients)

	Model 1 CPGE versus university and vocational higher education		Model 2 Vocational higher education versus university
	1962-1967 cohort	1975-1980 cohort	1962-1967 cohort	1975-1980 cohort
Father's occupation
Significance levels: *: p ≤ 0.05; **: p ≤ 0.01; ***: p ≤ 0.001.
Interpretation: the estimated coefficient for individuals defined by the active modality indicates the influence of that modality, all other things being equal, compared with the individuals defined by the reference modality.G2 (log likelihood) estimates the explanatory power of the model.
Source: authors' calculations from FQP 2003 survey.
Manual worker (ref.)	0	0	0	0
Farmer	–0.39	1.07 ***	0.86 ***	0.74*
Self-employed: business, trade and crafts	0.01	–0.14	–0.11	0.16
Higher-level occupation	0.30	0.38	–0.20	0.11
Intermediate occupation	–0.10	0.10	0.02	–0.10
Clerical or sales worker	–0.76 *	–0.21	0.18	–0.30
Other	0.00	0.14	–0.31	0.05
Father's qualification
CAP, BEP, BEPC (ref.)	0	0	0	0
No qualification or CEP	0.10	–0.11	–0.16	0.05
Baccalauréat	0.13	0.58 *	0.12	–0.02
Baccalauréat + 2 years	0.66 *	0.67 ***	–0.25	–0.12
Degree in higher education	0.74 ***	1.12 ***	–0.39 *	–0.35*
Type of baccalauréat
Vocational baccalauréat (ref.)	0	0	0	0
Literature or economics baccalauréat	0.99	1.30 ***	–1.45 ***	–2.14***
Science baccalauréat	2.78 ***	2.58 ***	–0.70 *	–1.47 ***
Technology	0.03	0.57	0.13	–0.22
Sex
Male (ref.)	0	0	0	0
Female	–0.58 ***	–0.46 ***	0.15	–0.11
Constant	–3.70 ***	–4.19 ***	1.07 ***	1.64 ***
G2	1,159	1,289	1,684	2,104
Sample size	2,111	2,454	1,341	1,733

appear to be correlated with family origin, except for a slight negative impact from the father's educational level and a more marked tendency among farmers' children to opt for vocational higher education rather than university. The greatest difference comes from the type of baccalauréat obtained: literature or economics baccalauréat-holders are by far the least numerous to opt for vocational higher education, followed by science baccalauréat-holders. For comparable baccalauréats, girls do not noticeably avoid vocational courses.

In conclusion, one may say that access to various types of higher education remains marked by social and gender inequalities, the former inequality most apparent in the link between the type of baccalauréat obtained and the choice of short vocational higher education courses. This is not the case for access to the elite sector; here, social origin (basically the father's educational level) plays a specific role in addition to the type of baccalauréat held: the odds ratio for having a science baccalauréat among general baccalauréat-holders is much lower than for access to the elite sector among general baccalauréat-holders. In terms of gender, the most significant observation is the lower probability among girls with a given type of baccalauréat of applying to the elite sector, rather than a reluctance to choose vocational courses. The "choice" pattern did not greatly change over the study period: whereas in general (the whole population), the chances of entering higher education (all sectors) have become more equal, the career choices of baccalauréat-holders have tended to become more inegalitarian, and in particular access to selective elite higher education, where no trend towards democratization is observed and odds ratios remain high (7.0) at the end of the period, markedly differing from those in other sectors.

Choosing a speciality, a horizontal choice?

In higher education, choices are finalized. It would be simplistic to consider them as unequal simply in terms of the prestige and salaries of the jobs to which they lead; they are also genuinely different. One may hypothesize that some areas of study appear more attractive to particular social groups because the children in those groups are familiar with certain values, activities, styles of work, etc. Some researchers (for example, van de Werfhorst et al., 2003) hypothesize that, in general, inequalities in higher education are increasingly "horizontal", and that, for example, law and medicine more specifically attract children from higher social categories, whereas teaching and engineering are more attractive for working-class children (and some sectors are apparently more "neutral", such as economics and natural sciences).

Should one consequently hypothesize, as do so many European researchers, that choices are mainly "horizontal" and not hierarchically ranked? We doubt it, because in explaining choices, it is difficult to distinguish between, on the one hand, genuine preferences for a given area of study and corresponding careers and, on the other, unequal initial advantages and the resulting unequal career prospects. Nevertheless, it is instructive to examine how students are spread across qualitatively different areas of study, this time without using the hierarchy of types of study. We isolated five major areas: law-economics-business, arts-humanities⁽¹⁹⁾, science-medicine-health and engineering.

Table 10.

Access of manual workers' children to various specialist subjects in higher education, compared with children of parents in higher-level occupations

	Percentage access			Difference between children of parents in higher-level occupations and manual workers' children
	All	Children of parents in higher-level occupations	Manual workers' children	Percentage point difference	Odds ratio
Source: authors' calculations from FQP 2003 survey.
Baccalauréat-holders
1962-1967 cohort
Arts, humanities	19.7	17.0	21.9	–4.9	0.7
Law, economics, business	31.8	28.4	33.8	–5.4	0.8
Sciences	16.3	23.1	9.9	13.2	2.7
Medicine, health	12.5	15.1	12.0	3.1	1.3
Engineering	17.9	15.8	20.7	–4.9	0.7
1975-1980 cohort
Arts, humanities	22.0	20.8	20.8	0.0	1.0
Law, economics, business	32.6	28.6	36.6	–8.0	0.7
Sciences	22.2	28.4	18.5	9.9	1.7
Medicine, health	8.1	10.0	6.1	3.9	1.7
Engineering	14.1	10.8	17.3	–6.5	0.6
Science baccalauréat-holders
1962-1967 cohort
Arts, humanities	5.7	5.5	5.0	0.5	1.1
Law, economics, business	14.9	13.8	8.4	5.4	1.7
Sciences	34.7	37.6	30.5	7.1	1.4
Medicine, health	24.3	23.6	32.4	–8.8	0.6
Engineering	19.5	19.4	22.3	–2.9	0.8
1975-1980 cohort
Arts, humanities	5.8	7.7	5.5	2.2	1.4
Law, economics, business	11.3	11.7	9.4	2.3	1.3
Sciences	50.5	49.7	50.2	–0.5	1.0
Medicine, health	15.4	16.8	11.5	5.3	1.6
Engineering	16.1	13.4	23.4	–10.0	0.5

If we ignore the level of study, and consequently the "vertical" aspect of the choice, by grouping in the health sector, for example, medicine and all paramedical courses, or, in the engineering sector, engineering schools, STSs and IUTs, the choices of young baccalauréat-holders appear to be relatively unmarked by their social origin and change little from one cohort to another. Admittedly, children of parents in higher-level occupations are over-represented in sciences and health, whereas manual workers' children are particularly numerous in engineering and law-economics-business. The only area where the two groups are equal is arts-humanities. But the odds ratios are fairly low: 1.7 for the two areas where there are more children of parents in higher-level occupations, and 0.6 and 0.7 for those where there are fewer.

In terms of specialities chosen, two significant changes may be observed from one cohort to the other. In health, both the short vocational courses and the longer ones are tending to become more socially selective, which is easily explained in the latter case given the increasingly selective nature of medicine, with the introduction of a numerus clausus in 1971. In sciences, there is a trend towards democratization, although this field is the second most socially selective after medicine. The other trends are less marked: both law-economics and engineering tend to attract rather more working-class students (perhaps because technology and vocational baccalauréats lead on more naturally to these subjects), whereas arts-humanities attracts rather fewer, with an odds ratio of 1.0 at the end of the study period.

The trends remain when science baccalauréat-holders are separated out, but the choices of these young people, who are the most selected academically, remain marked by sharp social differentiation (second part of Table 10). A clear distinction can be seen between the health sector, becoming more socially selective over time (as is arts-humanities, but with fewer numbers), and engineering, which attracts more working-class students and more so over time.

To observe these trends in greater detail, we estimated multinomial models for four major study areas (Table 11). Only the models estimated for cohort 2 (1975-1980) are presented here, but the comments point up the changes from cohort 1.

For law-economics-business (taken together) as opposed to arts-humanities, this choice is more often made in the first cohort by boys and technology baccalauréat-holders, and less often by baccalauréat-holders with good grades. Controlling for these factors, there is no social origin effect. By the second cohort, changes have occurred. The influence of the type of baccalauréat held is greater, with technology and vocational baccalauréat-holders more often opting for this subject area (probably more often business than the university disciplines of law and economics). Girls are still less likely to choose this subject area, and good grades are no longer associated negatively with this choice, which is therefore less often disparaged by young baccalauréat-holders of a good academic level.

In the case of sciences, this choice is made more in the first cohort by boys holding a general baccalauréat (but not necessarily a good grade), and social inequalities are noticeable. Again changes occur between cohorts: pupils with

Table 11.

Analysis of access to various subject areas in higher education, 1975-1980 cohort (coefficients of multinomial models)

	Law, economics versus arts-humanities			Sciences versus arts-humanities
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Significance levels: *: p . 0.05; **: p . 0.01; ***: p . 0.001.
Interpretation: Model 1 is estimated on basic variables (sex and social origin). Model 2 includes the type of baccalauréat variable and Model 3 the grade.
Source: authors' calculations from FQP 2003 survey.
Social origin
Non high-level occ. (ref.)	0	0	0	0	0	0
Higher-level occ.	0.18	0.46 **	0.46 **	0.36	0.35	0.30
Sex
Male (ref.)	0	0	0	0	0	0
Female	–0.69 ***	0.63 ***	–0.63 ***	–1.58 ***	–1.59 ***	–1.60 ***
Type of baccalauréat
General baccalauréat (ref.)		0	0		0	0
Vocational baccalauréat		1.63 ***	1.62 ***		–0.13	–0.23
Technology baccalauréat		1.46 ***	1.46 ***		–0.19	–0.18
Grade in baccalauréat
Pass grade (ref.)			0			0
High grade			0.08			0.48 ***
Constant	0.99 ***	0.41 **	0.39 *	0.95 ***	0.98 ***	0.83 ***
	Medicine, health versus arts, humanities			Engineering versus arts, humanities
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Social origin
Non higher-level occ. (ref.)	0	0	0	0	0	0
Higher level occ.	0.57 **	0.64 **	0.64 **	-0.30	0.11	0.08
Sex
Male (ref.)	0	0	0	0	0	0
Female	0.23	0.25	0.25	–3.07 ***	–3.00 ***	–3.00 ***
Type of baccalauréat
General baccalauréat (ref.)		0	0		0	0
Vocational baccalauréat		0.36	0.36		1.82 ***	1.74 ***
Technology baccalauréat	0.56	0.56		2.10 ***	2.12 ***
Grade in baccalauréat
Pass grade (ref.)			0			0
High grade			–0.01			0.39
Constant	–1.26 ***	–1.45 ***	–1.44 ***	1.24 ***	0.33	0.21
R2 = 0.152
Sample size = 1,123

Table 12.

Social origin of higher education graduates by subject area and type of course

	Percentage		Difference between children of parents in higher-level occupations and manual workers' children
	Children of parents in higher-level occupations	Manual workers' children	Percentage point difference	Odds ratio
Source: authors' calculations from FQP 2003 survey.
1962-1967 cohort
Arts, humanities
University (under-grad.)	21.3	22.9	–1.6	0.9
Elite and post-graduate courses	30.2	8.1	22.1	4.9
Law, economics, business
STS, IUT	16.8	23.9	–7.1	0.6
University (under-grad.)	23.7	21.3	2.4	1.1
Elite and post-graduate courses	40.1	4.4	35.7	14.5
Sciences
STS, IUT	22.9	17.6	5.3	1.4
University (under-grad.)	32.6	10.7	21.9	4.0
Elite and post-graduate courses	44.0	8.4	35.6	8.6
Health
STS, IUT	18.6	23.2	–4.6	0.8
University (under-grad.)	40.4	16.0	24.4	3.6
Elite and post-graduate courses	41.8	15.0	26.8	4.1
Engineering
STS, IUT	15.4	24.3	–8.9	0.6
Elite and post-graduate courses	46.8	15.1	31.7	4.9
1975-1980 cohort
Arts, humanities
University (under-grad.)	20.7	20.3	0.4	1.0
Elite and post-graduate courses	36.1	6.5	29.6	8.1
Law, economics, business
STS, IUT	17.6	30.7	–13.1	0.5
University (under-grad.)	29.4	19.2	10.2	1.8
Elite higher education and longer courses	41.0	8.9	32.1	7.1
Sciences
STS, IUT	19.5	24.4	–4.9	0.8
University (under-grad.)	28.1	16.7	11.4	1.9
Elite and post-graduate courses	50.7	13.1	37.6	6.8
Health
STS, IUT	21.4	17.0	4.4	1.3
University (under-grad.)	43.9	12.8	31.1	5.3
Elite and post-graduate courses	62.0	11.5	50.5	12.6
Engineering
STS, IUT	13.6	30.6	–17.0	0.4
University (under-grad.)	21.9	43.9	–22.0	0.4
Elite and post-graduate courses	30.5	9.4	21.1	4.2

high academic achievement are more attracted by this choice (good grades become significant) but the previously negative influence of holding a technology or vocational baccalauréat decreases. The influence of social origin is no longer significant, and science appears to have become the choice of good working-class pupils. On the other hand, girls remain reluctant to choose these subjects.

For the health sector (short and long courses taken together), in the first cohort, none of the characteristics examined (sex, social origin, type of baccalauréat, grades, etc.) has any significant influence. By the second cohort, higher social origin becomes a significant factor, which is consistent with the increased social selectiveness of this sector.

For engineering (short STS-IUT and long grande école courses together), this seems mainly to be the choice of boys with technology baccalauréats; neither good grades nor social origin have any significant influence. There is no change between cohorts; at most we note a higher impact for technology and vocational baccalauréats and the fact that girls are increasingly reluctant to make this choice.

Overall, if all choices are taken into account (which is possible with multinomial models), the type of baccalauréat held appears to matter in two main subject areas: law-economics-business and engineering. Social origin is influential in medicine and law-economics-business; good grades only count in sciences. Last, being a girl (rather than a boy) is negatively correlated with the choice of engineering and sciences and, to a lesser extent, with law-economics -business, and positively correlated with arts-humanities (there is no significant correlation with health careers).

Some of these results do not contradict the hypotheses of European sociologists, particularly the fact that engineering and sciences are the areas where social background is of least importance; in other words, where manual workers' children are least differently represented with respect to children of parents in higher-level occupations. Does this imply a "working-class preference" marked by parents' experience in industrial work (as van de Werfhorst suggests), while choosing the health sector is the result of close experience with parents in higher-level occupations? Our results do not support the idea that there are basically different preferences according to social origin; the odds ratios associated with these various subject areas are relatively low. Furthermore, the view that choices are "horizontal" and that preference outweighs ranking fails to explain certain contrasts: how can it be that the elite higher education courses are becoming socially more selective and engineering less so? It seems to us more appropriate, at least for France, to consider both the ranking of subjects and their areas. While preferences (and possibly differing preferences by social group) may direct a student towards a given subject area, the academic and socio-occupational advantages of students probably determine the type of course in which these preferences can be expressed (engineering school or BTS, medicine or paramedical course, etc.).

Clearly, if socially transmitted preferences were the key factor, these attractions should be observed at all levels of study; manual workers' children would be over-represented in all tracks leading to engineering, or children of parents in higher-level occupations in all those leading to the health sector. On the basis of our results, we can state that the differences expressed by odds ratios are greater between various levels of study than between subject areas. Ranking in the educational hierarchy would indeed appear to be more important than purely horizontal preferences (Table 12).

The highest odds ratios are almost always found for selective courses or post-graduate university courses, whatever the subject, while the socially most open courses are the short vocational ones, again whatever the subject. The figures confirm that the health sector has become more selective, particularly for longer medicine courses, but to a lesser extent also for other courses, whereas social selection has declined at all levels in sciences. The second cohort reveals the arrival of manual workers' children in university engineering courses that hardly existed at the start of the study period and have been extensively developed since then (particularly science and technology Master's degrees). At the end of the study period, this subject area is significantly more socially open than the others.

Underlying these observed changes and patterns of inequality – whereby course level is socially more selective than course subject – there are clearly strategies at work that must, one assumes, include career prospects. Without closely analysing in this paper the "return on investment" of various courses of study, we would simply note that in the light of our analyses of the FQP survey data, the chances of gaining a first job in a higher-level occupation are determined mainly by the type of qualification and only secondarily by the speciality. In this respect, the elite sector performs much better than the universities and short vocational courses. It is easy to understand why social competition mainly concerns access to the grandes écoles and the courses that prepare for them.

Conclusion

This analysis of changes in inequality in higher education over the period of rapid expansion of secondary education between 1985 and 1995 confirms that quantitative opening-up is not a sufficient condition in itself for reducing inequality. The undeniable democratization of the baccalauréat (an equalizing democratization, to use Merle's expression) has led to a "uniform" democratization in access to higher education, whereby differences remain unchanged. In addition there is a segregative democratization in access to the baccalauréat, with "new baccalauréat-holders" from the working class being recruited from among students with a BEP qualification that leads on to a vocational baccalauréat. Given the close ties between secondary and higher education in France, this phenomenon operates as a restriction on the "potential return on investment" from the baccalauréat in higher education. There have also been changes in supply and demand, with a stability in the relative proportion of the elite sector (an understandable Malthusian approach if career opportunities are to be maintained), which makes competition increasingly fierce. In addition to these phenomena that restrict entry, there are also the strategies adopted by young people themselves, especially those who have the most advantages and wish to stay ahead.

Higher education appears to be following a traditional pattern whereby a dam is breached at one level, and defence shifts to a higher level, until this dam too gives way. Opening year 6 to all was followed by greater selectiveness in the college; removing track choice at the end of year 7 was followed by more selective orientation at the end of year 9 and then of year 10. The "flood wave" has now reached the gates of the university. But here, young people are supposed to aim for a final degree so as to begin a career. The process cannot go on for ever, simply displacing inequality further up the system, because at some stage there has to be a connection with actual employment, which will not necessarily change to adapt to the incoming flow of graduates.

The limitations of this educational policy (not excluding individual strategies) that merely shifts educational inequality without naturally taking account of the jobs that will follow, are evident. The result is a relative devaluation of qualifications and a scissor effect described by Forsé (1997, see also Chauvel, 1998, and Duru-Bellat, 2006): manual workers' children more often pass the baccalauréat, but the type of baccalauréat they pass and the type of higher education to which they have access brings them onto the labour market with qualifications that are very likely to channel them into manual or clerical occupations. At the other end of the spectrum, it is easy to understand why children of parents in higher-level occupations focus on the elite sector, which has managed to maintain its value on the labour market and to guarantee access to a high-level career. Beyond analyses of the varying preferences of children from given social backgrounds and the increase in horizontal differentiation, there is indeed competition for unequal jobs, which begins very early, since by the end of year 9, the pupils with the most difficulties (many of whom have been struggling since primary school) are directed towards short technology courses that lead on to a vocational baccalauréat, while others move up into year 10 with options enabling them to aim for a general baccalauréat.

Promoting equal opportunity cannot be restricted to making marginal changes to the selection process for the elite sector via various systems of positive discrimination (such as special admission procedures for pupils from disadvantaged lycées) when, by definition, this sector will always be quantitatively limited. Action needs to be taken far upstream, to address inequalities in early academic success and the opportunity for entering the tracks most suited to the elite sector (in this case, access to general baccalauréats). Many forms of downstream action are naturally conceivable, such as encouraging life-long learning, which is particularly lacking in France, and whose absence tends to compound early inequalities established at the start of the educational trajectory.