Downward Social Mobility Across Generations: The Role of Parental Mobility and Education

by Susanne Alm
Institute for Futures Studies

Sociological Research Online, 16 (3) 2

Received: 24 Sep 2010     Accepted: 9 Jul 2011    Published: 31 Aug 2011


Intergenerational downward social mobility is an issue of growing relevance, but there are still very few studies examining possible risk factors for dropping down the occupational hierarchy. On the basis of unique longitudinal interview and register data from Sweden, this study analyses the roles played by parental upward mobility and parental levels of education in downward mobility. Elements from cultural capital theory (CCT) are investigated as possible mechanisms for explaining the relationship between independent and dependent factors. Whereas the study fails to find support for the role of parental mobility, the parents' level of education turns out to be a powerful predictor of downward mobility. And whereas the measure of cultural capital presents a weak or non-existent relationship with the dependent variable, two attitudinal variables, employed as indicators of habitus and a possible don Quixote effect, do present a significant relationship with the risk for downward mobility. However, while the study hypothesized these attitudes to be mediating mechanisms that might explain the relationship between parental educational level and downward mobility, in the multivariate analyses the attitudinal measures instead turn out to have an independent effect in addition to the parental level of education.

Keywords: Intergenerational Downward Mobility, Parental Mobility, Parental Educational Level, Cultural Capital Theory, Longitudinal Study


1.1 Downward social mobility is an issue of growing relevance. The relative increase in the number of more privileged positions on the labour market and the rapid expansion in educational opportunities witnessed in western countries over the post war decades have both slowed down in recent decades, putting an end to demand-driven upward mobility and instead inducing a shift towards a symmetric intergenerational fluidity between social classes (e.g. Goldthorpe and Jackson 2007; Tolsma et al. 2009; Deary et al. 2005). Since intergenerational upward social mobility is typically considered a corner stone of democratic and open societies (e.g. Sellers 2000), this development raises interesting questions and challenges. Are we ready to face a situation where downward mobility comes to constitute a precondition for the possibility of moving upwards through the hierarchy?

1.2 At the individual level, these developments give rise to the question of who is at risk of being downwardly mobile? Although social mobility is a central research area for sociologists, knowledge on which individuals become downwardly mobile remains scarce. At the same time, the avoidance of downward mobility has been postulated as the guiding principle of one of today's most influential theories on social inequality in educational outcomes – the relative risk aversion theory - RRA (Boudon 1974; Breen and Goldthorpe 1997; Goldthorpe 1996; Breen 2001). In the context of empirical testing, RRA has received varying support. While van der Werfhorst and Hofstede (2007) and Becker and Hecken (2009) do find support for the theoretical propositions of RRA, no such support has been found in studies by Stocké (2007) and Gabay-Egozi et al. (2009). In the current article, the theoretical assumptions of RRA are merely used as a point of departure for the analysis: if we assume that the avoidance of downward mobility is the most central concern for individuals when making educational choices, how is it that some individuals still drop down the social hierarchy? To address this question, the article employs a theoretical framework based on previous research conducted into both downward mobility and educational inequalities. We explore the role of upward mobility in previous generations and that of parental levels of education for the risk of becoming downwardly mobile (see Jackson and Marsden 1962; Richardson 1977), and we employ elements from cultural capital theory, CCT (Bourdieu 1984; Bourdieu and Passeron 1977; Bourdieu 1986) to develop the theoretical perspective. Our hypotheses are that upward mobility in previous generations and lower levels of parental educational achievement respectively both increase the risk for downward mobility (cf. Richardson 1977). The hypotheses highlight differences in the social background of individuals, which means that CCT may be of help in interpreting the results. Whereas previous research on the topic has only dealt with short-distance mobility across the white – blue-collar line (Richardson 1977), this article studies downward mobility from higher white-collar to both lower white-collar and blue-collar positions.

1.3 The analyses are based on a unique longitudinal database, the Stockholm Birth Cohort Study (Stenberg et al. 2006; Stenberg and Vågerö 2005), in which individuals can be followed from childhood until middle age. In addition to containing information on the subjects themselves, the database also includes data from interviews with (a sample of) their parents.

1.4 The next section presents a review of previous research on downward mobility and educational inequality. The article then describes the SBC database, followed by the operationalizations employed in the study, before moving on to present and analyse the findings. The final section of the article presents a summary and a number of concluding remarks.

Previous studies of downward mobility

2.1 While intergenerational upward social mobility is often viewed as both a 'natural' and desirable phenomenon within an open and democratic society, downward mobility has often been regarded as something 'unnatural' and as the result of individual failure (Sellers 2000). In part this may be due to the low rates of downward mobility we have previously experienced – in a situation where only a small proportion of individuals become downwardly mobile, more vulnerable individuals are likely to constitute a larger proportion of those affected. One way of describing this phenomenon would be to say that the rate of downward mobility has a compositional effect on the group that is downwardly mobile (see Alm 2001 for a similar discussion concerning unemployment). Probably for the same reasons, significantly less research has been focused on explanations for and the consequences of downward mobility within the class hierarchy, than on upward mobility. However, as has recently been shown by Goldthorpe and Jackson (2007), at least for men, rates of downward mobility have increased in recent years and structural developments indicate that this trend is likely to continue.

2.2 The relatively few studies that have been conducted into downward mobility have typically focused on different types of social problems, either in the family of origin or on the part of the individual him- or herself (e.g. West 1991, Rodgers and Mann 1993, Timms 1998, Hemmingsson et al. 1999.) Richardson (1977) has referred to this understanding of downward mobility as the social casualty perspective. Mental illness, criminality and alcoholism and/or drug abuse have all been proposed as explanations for why individuals slide down the occupational hierarchy. In a previous study based on the same data material as that employed in the current study (Alm 2008), a correlation was found between registered drug abuse prior to age 30 and downward mobility, albeit only for boys. However, as has been noted by Parkin (1971), the social casualty theory is perhaps more applicable to cases where an individual drops through the entire occupational hierarchy to finish up with non-existent or only very weak ties to the labour market. Further, in line with the above discussion on the presence of a compositional effect (Alm 2001), social problems ought to have a greater impact in relative terms when there is substantial upward pressure within the occupational hierarchy, when the total proportion of downwardly mobile individuals is small, and where downward mobility is thus the result of a more powerful selection process.

2.3 Attempts have also been made at explaining downward mobility in terms of other factors, however, and an interesting question is that of whether theories that explain upward mobility can be inverted in order to learn more about downward mobility. This is Richardson's (1977) belief when he draws upon a study of Jackson and Marsden (1962), in which upward mobility is in part explained by 'temporary' downward mobility in the previous generation. Thus Richardson (1977) argues that downward mobility may be explained by reference to a temporary upward mobility in the previous generation. The downwardly mobile individual ends up in a position similar to that of his or her grandparents, after a 'temporary' upward movement in the individual's parents' generation. In this way, Richardson argues, downward mobility results in 'status consolidation' rather than in 'status loss' (p. 304). However, there is something of a slide in the focus of Richardson's argumentation, since he later states that not all mobility can be explained in this way, but only that where the upward mobility of the parents did not occur in relation to education. The reason for this is found in the role played by values and norms in the context of occupational (and educational choices). In Richardson's (1977: 305) words:

'...[N]ot all upward occupational mobility is necessarily translated into social mobility in the sense of it involving normative and relational shifts as well as economic…. The crucial variable would seem to be whether the mobility proceeds through a formal educational route and is therefore "legitimate" or whether it proceeds via a non-educational mobility route and is therefore "illegitimate". In the latter case…the lifestyle, attitudes and social ties of these upwardly mobiles are likely to be at odds with their destination class.'

2.4 The idea is thus that in families where one or both parents have been upwardly mobile with respect to occupation but not with respect to education, the norms and values associated with their social origins remain rather unaffected by this mobility.[1] And since these norms and values are internalized by the child in the socialization process, downward mobility is expected to occur disproportionately often within these families.[2]

2.5 Although both factors seem relevant to the study of downward mobility, we would argue that they should be studied separately. We thus analyse downward mobility in relation to mobility in previous generations and parental education respectively, and our hypotheses are that parental upward mobility and a low level of parental education are both related to a higher risk of downward mobility for our study subjects. We also combine the two factors in search of an interaction effect. By comparison with Richardson's study we have a larger number of observations and hence better opportunities for the use of multivariate analysis techniques.[3]

Theories of social inequality in education

3.1 As has been noted by Richardson (1977), one problem with many mobility studies is that they (implicitly) view the social origin of mobile individuals as fixed and as an irrevocable given. In this regard, Richardson's perspective offers an interesting exception. Taking mobility on the part of the parents into consideration also provides an opportunity to apply theories of educational inequality to social mobility. While theories of educational inequality aim to explain why individuals from different social backgrounds proceed to higher education to a varying extent, studies of which individuals become socially mobile naturally deal with individuals from similar social origins. However, if information about mobility is available not only for the individuals themselves, but also for their parents, there are observable differences in social background and thereby possibilities for comparison.

3.2 The RRA perspective is difficult to apply to a situation without variation in parents' occupational position (of destination) – since it is precisely this variation which, according to the theory, yields different educational aspirations among individuals from different social classes (e.g. Boudon 1974; Breen and Goldthorpe 1997; Goldthorpe 1996; Breen 2001). However, the other main school of thought when it comes to educational inequality, i.e. cultural capital theory, CCT (Bourdieu 1984; Bourdieu and Passeron 1977; Bourdieu 1986) is better suited to such situations.

3.3 According to Bourdieu, classes differ with respect to the amounts of capital – economic, cultural and social – that they control. Bourdieu revolutionized stratification research by placing the concept of cultural capital in focus, and by recognizing its vital role as a power resource. In addition to quantitative differences in terms of resources, according to Bourdieu each social class also has its distinctive habitus - a system of predispositions such as values and motivations, or if you wish, a class-specific way of seeing the world which affects experiences in all areas of society.

3.4 An individual's habitus is affected by social mobility, but changes are slow. Bourdieu (1984) introduces the concept of the don Quixote effect to describe situations where an individual's habitus has not (yet) become aligned with his or her new social position, something which may express itself in e.g. attitudes, language use or manners and which produces the impression of the individual being out of place in a given situation.[4]

3.5 To control cultural capital means to know the dominant culture in society. It means knowing cultural codes on behaviour and language use and it is inherited by children from their parents. Since the dominant culture is the culture expressed in school, children from families with large amounts of cultural capital will tend to do better in school and to proceed to higher education to a greater extent than others. Alternative ways of formulating the nature of this inequality are to say that the educational system demands knowledge of a kind that is only possessed by those from more privileged families (e.g. Sullivan 2001), or that student performance is not evaluated in accordance with class-neutral standards (Barone 2006).

3.6 The current study makes use of CCT to specify the way in which mobility in previous generations, and the educational level of parents, can each contribute to the explanation of downward mobility. Are there signs of what could be termed a don Quixote effect among parents who have themselves been mobile? And is there a relationship between on the one hand the educational level of the parents and on the other attitudes towards theoretical knowledge and preferences concerning the future occupation of the child? Further, do we find a difference in the level of access to cultural capital between children whose parents have and have not been socially mobile and between children whose parents have different educational levels? Cultural capital, attitudes towards theoretical knowledge and preferences concerning children's future occupation are thus seen as possible mechanisms for explaining relationships between the independent and dependent factors examined by the study, i.e. parental mobility and parental education on the one hand and downward mobility among the study subjects on the other. Where this is motivated, we also have the opportunity to control for the so-called primary effects of educational inequality (Boudon 1974).[5]

The Stockholm birth cohort study

4.1 The SBC is a longitudinal database created by means of a fusion of two anonymised data sets. The first is the Metropolitan Study, which comprises all individuals born in 1953 and resident in Greater Stockholm ten years later (Jansson 1995). The Metropolitan Study includes a massive amount of register and survey data about the individuals in the sample themselves as well as their parents.

4.2 The second data set is the Health, Illness, Income and Employment database (the HSIA). This comprises register data on all individuals living in Sweden in 1980 or 1990, and it includes information on amongst other things income, occupation and welfare benefit recipiency.

4.3 Since both databases have been anonymised, a probability matching process was employed, as a result of which 96 percent of the observations – 14,294 individuals – were able to be matched.[6] By combining the two databases it is possible to follow the individuals born in 1953 who lived in Greater Stockholm at the age of ten until they are 48 years old.

4.4 The data employed in the current study are drawn from several sources, with register data from the HSIA and survey data from what have become known as the School Survey and the Family Study constituting the most important of these. In 1966, when the individuals in the study were aged 12–13, they completed a large survey questionnaire in school, which then became known as the School Survey, answering questions on amongst other things their attitudes to school, how much they enjoyed being in school, their leisure time activities and their plans for the future. Two years later, in 1968, a sample of parents completed a large questionnaire which included questions on amongst other things their own childhood, and on child-rearing and attitudes towards education. This survey, which was completed in the vast majority of cases by the mothers, became known as the Family Study. Approximately 4,000 mothers were included in the sample and the response rate was 91 percent, yielding 3,651 observations. Since the current study uses information from the parents, 3,651 is our total number of observations. It should of course be noted that results based on analyses of the SBC material may not necessarily be generalizable to individuals born at other times and living in other parts of the country.


Downward mobility

5.1 As was noted earlier, Richardson only deals with what he terms 'middle mass mobility', i.e. short-distance mobility across the white-blue collar line. According to in-depth interviews employed as a complement to Richardson's quantitative analyses, the parents of the downwardly mobile respondents had often taken less comprehensive steps across the mobility line. Further, these steps had often been taken at somewhat higher ages and above all, as noted earlier, they involved what would be termed 'working one's way up' rather than mobility based on educational achievement. The relatively small moves that had been made by the parents resulted in many of the respondents actually not perceiving themselves as being downwardly socially mobile. This is in line with the arguments made by researchers over recent years that lower white-collar positions are in practice more similar to blue-collar positions than they are to intermediate and higher white-collar positions. For example, in the Goldthorpe class schema, as a result of similarities in employment contracts, the category of routine non-manual workers is usually collapsed with those of manual workers (e.g. Goldthorpe and Jackson 2007). The same argument has been put forward by Ahrne in relation to the Swedish context (e.g. Ahrne et al. 1996) and in the Swedish classification system, lower white-collar positions are often collapsed with blue-collar positions. As was mentioned above, the current study instead investigates the explanatory potential of Richardson's hypothesis for longer distance mobility.

5.2 The focus of the study is directed at mobility from higher white-collar to lower white-collar and blue-collar positions. Since we cannot assume the same explanations of downward mobility to be relevant for mobility from different levels (Richardson 1977), it is preferable to keep the occupational position of the parents (and thereby the class of origin of the study subjects) as homogenous as possible.[7] In addition to the theoretical arguments, there are also practical reasons for using this design, namely that occupational classification systems have shifted during the period covered by the study and it has not been possible to access the original questionnaires in order to recode the occupational data. Since the 1970s, the most commonly used socio-economic classification system in Sweden has been the SEI (Statistics Sweden, 1982). The SEI is based on occupations and categorises individuals according to the length of post-comprehensive education required. It also distinguishes between self-employed and employees, between employees with and without subordinates and according to trade-union affiliation. The SEI-categorisation is very similar to the classes specified by Goldthorpe (Erikson and Goldthorpe 1992). At the time of the School- and Family Studies however, the SEI had not yet been introduced and socio-economic classifications were still made with reference to the so called, 'Social group classification system', which was also compiled by Statistics Sweden, but as long ago as 1911 for the purpose of electoral studies (SOFI 2005). Table a in the appendix presents the distribution for (dichotomized) social class in the three generations analyzed in the study. It can be seen from the table that the proportion of individuals in higher white-collar positions is lower in the parental generation than among both the grandparents and the study subjects themselves. This is an effect both of the change in classification systems between the two older generations and the relative expansion in higher positions in the study subjects' generation. If anything, the slight difference in proportions means there is a tendency to underestimate the proportion of upwardly mobile individuals in the parental generation.

5.3 The difficulties associated with comparisons between the two classification systems are primarily related to intermediate levels. The solution to the problem employed in this study was to select only those subjects whose parents had a high white-collar position in the 'Social group classification system' (self-employed excluded)[8]. This then means that the individuals classified as downwardly mobile are those who have lower white-collar and blue-collar positions based on the SEI classification[9], whereas the remaining individuals are considered to have consolidated their parents high position.[10] The occupational position of the respondent's family of origin is coded in accordance with the dominance method (Erikson 1984).

5.4 Occupational information for the parents is drawn from the Family Study conducted in 1968, whereas the same information for the study subjects is based on register data from 1990, i.e. when the subjects were 37 years of age. At this age, careers can be expected to be relatively mature, which is preferable (Deary et al. 2005). Out of the 3,651 individuals included in the Family Study, 733 (20 percent) had parents with a high occupational position.

Independent variables

Parental mobility

5.5 The subjects' parents are classified as upwardly mobile if their parents (i.e. the subjects' grandparents) held lower white-collar or blue-collar occupations. For the parents to be classified as upwardly mobile requires that none of the grandparents held middle- or upper white-collar positions. Parental level of education

5.6 Due to the rapid expansion of higher education in the second half of the 20th century (e.g. Brown 1995; van der Werfhorst and Andersen 2005), which was not matched by a similar growth in the number of higher-level positions on the labour market, western countries have over recent decades faced what has been termed credential inflation (van der Werfhorst and Andersen 2005). As a consequence, the correlation between educational level and occupation has weakened somewhat (van der Werfhorst and Andersen 2005). One practical consequence of this trend is that it complicates educational comparisons across generations – a 'high' level of education is a relative concept, particularly over time. In the SBC data, for example, around 50 percent of the subjects have a certificate of further education[11], whereas the corresponding proportions are 25 percent for the subjects' parents and only 10 percent for their grandparents. In the current study, the certificate of further education is considered to represent a high level of parental education. The parental level of education has also been coded according to the dominance method (Erikson 1984).


Cultural capital

5.7 Bourdieu is not very precise about how cultural capital should be measured in order to capture the advantages enjoyed by children from privileged families in the educational system (Sullivan 2001) and in part as a result of this vagueness, researchers have presented a variety of operationalisations of this concept (e.g. di Maggio 1982; Jonsson 1987; Crook 1997; de Graaf et al. 2000). Some measures have proven better than others however. For example, both Crook (1997), de Graaf et al. (2000) and Sullivan (2001) find that reading is associated with academic success, whereas participation in beaux-arts, i.e. in formal cultural activities outside the home, such as visits to museums and to the theatre, is not. One interesting interpretation of this finding is that whereas participation in formal culture primarily serves to communicate status, private cultural consumption, such as reading books, is a means of intellectual development (Crook 1997; de Graaf et al. 2000).[12] The current study employs a measure of cultural capital that is closely related to reading, namely one which asks about the number of books available in the home.[13] The question (posed to the parents in the Family Study) asks: How many books do you think there are in your home? There were nine response categories, from 'None' to 'About 3000 or more'. The nine categories were here collapsed into four: 1) Up to about 100, 2) About 300, 3) About 1000, 4) About 3000 or more. Besides being included in the SBC, this question has also been used in e.g. the Swedish Level of Living Surveys (e.g. Fritzell and Lundberg 1994).[14] Attitudes towards theoretical knowledge and preferences concerning the child's future occupation

5.8 Parental attitudes to theoretical knowledge and their preferences concerning their child's future occupation were investigated as indicators of habitus and hence of a possible don Quixote effect. The question used to measure attitudes to theoretical knowledge asked: Do schools place too much emphasis on theoretical knowledge and too little on practical skills? There were five response categories, from 'Yes, definitely' to 'No, definitely not'. The five response categories were here collapsed into three.[15]

5.9 Parental preferences concerning the future occupation of their child are measured on the basis of a direct question posed to the parents in the Family Study as to what occupation they would prefer their child to aim for. Apart from a direct question posed to the parent who completed the questionnaire (usually the mother) about her (or his) preferences, she (or he) was also asked to state what she (or he) believed the preference of the spouse to be. Answers were coded according to the occupational classification scheme employed in the study (and discussed above). The variable employed in the analysis distinguishes between 1) Families where at least one of the parents stated a preference for the child to aim for a higher or intermediate white-collar position, 2) Families where neither of the parents stated a preference for a higher or intermediate white-collar position, but where at least one of them preferred a lower white-collar or a blue-collar position, and, 3) Families where neither of the parents expressed any preference concerning the child's future occupation.

5.10 Since it is possible that parental preferences concerning a child's future occupation (i.e. in Boudon's terms secondary effects) are influenced by the child's academic ability (i.e. primary effects), the multivariate analyses include a measure of academic ability in the models. This measure consists of the score on an ability test distributed to the children at twelve years of age. The test included three sections: numeric, verbal and spatial, and the maximum score was 120 points (40 points on each section) (SOFI 2005).


Bivariate analyses

Parental mobility and parental level of educational 6.1 Of the 733 individuals from a privileged social class origin, there were 60 whose social class in adult life could not classified. Of the remaining 673 individuals, 463 or 69 percent have consolidated their parents' occupational (class) position, while the remaining 31 percent (210 individuals) have been downwardly mobile. Forty-six percent of the downwardly mobile individuals are males and 54 percent are females - there is thus a small over-risk for downward mobility for females as compared to males.

6.2 Among the 673 individuals with a privileged class background, 285 individuals (42 percent) have grandparents with a low social class position. In the Family Study sample as a whole, 70 percent of the subjects have grandparents with less privileged class positions of this kind. These figures point to a strong correlation between the social class positions of different generations, or in other words, to a low level of intergenerational fluidity between social classes.

6.3 Table 1 presents a cross-tabulation of downward mobility among the study subjects and the social class position of their grandparents, i.e. social mobility in the parental generation. There is a tendency in the expected direction, but the difference does not quite reach significance at five percent level (p= .06). Of those whose parents had been upwardly mobile, 35 percent have become downwardly mobile, as compared with 28 percent of those whose parents had consolidated a high occupational position.

6.4 As we saw, the in-depth interviews conducted in Richardson's study revealed that the relationship between parental upward mobility and subsequent downward mobility among their children only seemed to be applicable to cases where the upward mobility of the parents had no educational basis. This raised the question of whether the educational level of parents is in fact a better predictor of downward (occupational) mobility than parental mobility overall. To begin with, therefore, we investigated the role played in downward mobility by the parents' level of education alone.

6.5 Table 2 presents a cross-tabulation of downward mobility among the subjects and their parents' level of educational. In the Family Study as a whole, 26 percent of the subjects' parents have a high level of education. The corresponding proportion among those from a privileged class background is 82 percent, indicating that there is a strong correlation between educational level and occupational position. From Table 2, we can see that there is a very clear relationship in the expected direction, in that 51 percent of the children whose parents have a low level of education have been downwardly mobile, as compared to only 27 percent of the children whose parents have a high level of education (p < .000). Thus, the level of parental education appears to be a rather powerful predictor of downward intergenerational mobility.

6.6 The two factors were then combined in a manner (implicitly) proposed by Richardson. The results of this combination of the two factors are presented in Table 3. In line with our expectations, the smallest proportion of downwardly mobile individuals is found among those whose parents have a high level of education and whose grandparents also had a high occupational position. However, the educational level of the parents is of much greater significance in this context than the occupational position of the grandparents. Among those whose parents have a low level of education, the group with the highest proportion of downwardly mobile individuals is, surprisingly and in stark contrast to our expectations, comprised of those whose grandparents had a high occupational position. It should be noted, however, that this group is small (n=38) and the most reasonable interpretation of the analysis is that the educational level of the parents constitutes the only factor that is clearly of relevance for the risk of downward mobility in our sample. At the same time, since we have noted that there was at least a suggestion that parental mobility might also play a role (Table 1), this was also included in the subsequent analyses.

Cultural capital, attitudes to theoretical knowledge and preferences concerning the child's future occupation

6.7 Tables 4–5 below shows cross-tabulations between cultural capital, attitudes to theoretical knowledge and preferences concerning the child's future occupation on the one hand, and parental upward mobility and education on the other.

6.8 There is a strong relationship in the expected direction between cultural capital and parental upward mobility. Families where respondents have been upwardly mobile tend to control less cultural capital than families where the parents have consolidated the high occupational position of their parents. For example, in families where parents have consolidated a high occupational position, 35 percent claim to have only up to about 300 books at home, whereas the corresponding proportion in families where parents have been mobile is 44 percent.

6.9 There is also a significant correlation in the expected direction as regards attitudes towards theoretical knowledge, albeit not quite as strong as that found in relation to cultural capital. While 27 percent of parents who have consolidated a high occupational position strongly agree with the statement that school places too much emphasis on theoretical knowledge and too little on practical skills, the corresponding proportion among parents who have been upwardly mobile is 35 percent. These results could be interpreted in terms of a don Quixote effect.

6.10 Finally, with regard to preferences concerning the child's future occupation, only 50 percent of the parents stated any precise preferences. Of these, a majority stated a preference for a higher white-collar position. However, the only significant difference with respect to parental mobility is that parents who have been mobile are more likely than parents who have consolidated a high occupational position to state a preferred occupation of any kind.

6.11 The same bivariate analyses were run in relation to the educational level of the parents and the results of these analyses are presented in Table 5. As regards cultural capital, the results are striking. Whereas only four (4) percent of parents with a high level of education stated that there were at most about 100 books in their homes, the corresponding proportion among those with a low level of education is 30 percent. And whereas 63 percent of parents with a high level of education estimated the number of books in the home to be at least 1000, only 24 percent of parents with a low level of education stated that this was the case.

6.12 As regards attitudes to theoretical knowledge, the results are also clearly significant and point in the expected direction. Thirty percent of the parents with a high level of education disagreed with the statement that school places too much emphasis on theoretical knowledge and too little on practical skills, as compared to fifteen percent of the parents with a low level of education.

6.13 And finally, with regard to preferences concerning the child's future occupation, the results are once again significant and point in the expected direction. Thirty-three percent of the parents with a high level of education expressed a preference for their child to aim for a high or intermediate white-collar position, as compared to 22 percent of the parents with a lower level of education. (It should be remembered that around 50 percent of the parents declined to state a preference).

Multivariate analyses

6.14 The independent measures, together with the hypothesized mechanisms, were next combined in a logistic regression analysis of downward mobility. To begin with, as can be seen from Model 1 of Table 6, the model only includes the parental level of education and parental mobility, together with a control for sex. The effect of parental mobility on downward mobility, which in the bivariate analyses was quite weak but significant at the 10 percent level, is weak and does not reach significance. By contrast, the parents' level of education stands out as a powerful predictor of downward mobility. The tendency for girls to be downwardly mobile to a greater extent than boys is not significant.

6.15 In Model 2 of Table 6, the study's hypothesized mechanisms, i.e. the measures of cultural capital and habitus, are included, while the measures of parental education and mobility are temporarily left out. Overall, the explanatory power of the hypothesized mechanisms is quite strong, but there are variations between the factors. The weakest of the factors investigated is the measure of cultural capital. Although estimates are in the expected direction, they are quite weak and the only significant difference found is that between those households with the fewest books and those with the largest number of books (p= .05). The measures intended to capture aspects of (parental) habitus present stronger correlations with the risk for downward mobility. Firstly, the children of parents with very positive attitudes to theoretical knowledge are significantly less likely to be downwardly mobile (p= .004). And secondly, the children of parents who would prefer them to aim for a higher white-collar position are more likely to do so by comparison with the children of parents whose preference was for them to aim for a lower white- or a blue-collar position (p< .000). An interesting result is that the children whose parents declined to state a preferred occupation are also less likely to become downwardly mobile than the children whose parents stated preferences for lower white- or blue-collar occupations (p= .045).

6.16 The third model includes both the parental level of education and parental mobility together with the measures of cultural capital and habitus (Model 3). The study's hypotheses would be supported by the data if the indicators of cultural capital and habitus remain significant and if at the same time, the estimate for the parents' level of education (which is the only one of the two independent variables with a significant effect) is weakened. However, this is not what Model 3 shows. Two of the three expected mechanisms remain significant, but the effect of the parents' educational level is not notably affected by their inclusion. Thus the model indicates that the expected mechanisms do play a role in the risk of downward mobility, although not in the form of mechanisms that might explain the relationship between downward mobility and the parents' level of education, but rather by producing independent effects in addition to the effect of the parental educational level. However, this is only true in relation to the two attitudinal indicators, and the measure of cultural capital does not contribute to our understanding of downward mobility in the context of this model. The indicators of attitudes to theoretical knowledge and of parental preferences concerning the child's future occupation are not seriously affected by the introduction into the model of the study's principal independent variables.

6.17 As has been noted above, it is likely that parental preferences concerning their children's future occupation are affected by the children's academic ability and performance. Or, to use the words of the Boudon school, in order to arrive at a fair estimation of secondary effects, it is necessary to control for primary effects. Therefore, in the final model of Table 6 we also introduce a measure of ability consisting of the results from a test distributed to the respondents at twelve years of age. Not surprisingly, there is a strong relationship between academic ability measured in this way and the risk of downward mobility. Both the measure of attitudes to theoretical knowledge, and most importantly, the measure of parental preferences about the child's future occupation are also weakened by the introduction of the ability measure. But whereas the measure of attitudes to theoretical knowledge variable is in this model only significant at the ten percent level (p= .08), the independent effect of parental preferences in relation to the child's choice of occupation choice remains significant at the five percent level (p= .045). The independent effect of the parents' level of education is also somewhat weakened by the introduction of academic ability into the model, but parental education remains a powerful predictor of downward mobility.

Summary and concluding remarks

7.1 Downward social mobility is a factor of growing relevance in many western countries, but research on explanatory factors remains scarce. Building on a study by Richardson (1977), this study set out to investigate the roles played by parental mobility and parental level of education in the risk for downward mobility in a Swedish cohort born in 1953. Our hypotheses were that (upward) parental mobility and a low level of education among the parents would both be related to a higher risk for downward mobility among the study subjects. A large sample and unique longitudinal data have made it possible to conduct multivariate analyses on the basis of information relating to three generations. The study used CCT to develop the theoretical perspective and it was hypothesized that indicators of cultural capital and habitus might serve as mediating factors between parental mobility and educational level on the one hand, and the downward social mobility of the study subjects on the other.

7.2 Whereas multivariate analysis showed only a weak and non-significant relationship between (upward) parental mobility and downward mobility among the subjects (their children), the educational level of the parents proved to be a very powerful predictor of downward mobility. Thus the children of parents with low levels of education (despite their high occupational positions) ran a considerably higher risk of downward mobility than the children of parents with similar occupational positions but with a higher level of education. Cultural capital was measured on the basis of an approximation of the number of books available in the household (of the subjects' parents). In the multivariate analyses, this indicator presented only a weak or a non-existent relationship with the dependent variable. We cannot say definitively whether this result means that cultural capital truly plays no role in the context under study, or whether it is rather due to the indicator not being suitable for our purposes. However, among the variety of measures researchers have used to capture culture capital, reading has shown itself to be one of the better ones in predicting academic success (e.g. Crook 1997; de Graaf et al. 2000; Sullivan 2001). And it seems reasonable to expect there to be a fairly strong correlation between reading and the possession of books.

7.3 The indicators of habitus and of a possible don Quixote effect did however present strong and significant correlations with the risk for downward mobility. Thus, the children of parents with a positive attitude towards theoretical knowledge run a lower risk of downward mobility. Likewise, children whose parents state that they would like their child to aim for higher or intermediate white-collar positions, run a lower risk of downward mobility than children whose parents would prefer their child to aim for a lower white-collar or a blue-collar position. However, these attitudes did not mediate the effects of the parental level of education on downward mobility as was expected, but they rather turned out to be independent factors in relation to the dependent variable. Thus the study failed to find support for the hypothesis that CCT might help explain the relationship between parental levels of education and downward mobility.

7.4 It is difficult to come up with any concrete reasons why the study's result should not be generalizable to other western countries. Macro-level comparative studies have shown Sweden to be characterised by a higher level of social fluidity by comparison with most other western countries (e.g. Breen and Jonsson 2007). The process of educational equalization has been long and gradual, and had produced notable effects as early as the first post war decades (Breen and Jonsson 2007). Since educational reforms similar to those in Sweden have also been implemented in many other countries, a key question for researchers is that of how the case of Swedish exceptionalism should be understood. Although the mechanisms involved have yet to be specified, Breen and Luijkx (2004) have found that a considerably larger part of the total relationship between origins and destination is mediated via education in Sweden than in other countries. This might perhaps in turn be explained by Sweden's traditionally low levels of income inequality, which may contribute to relatively homogenous living conditions and lifestyles and leave less room for the effects of other criteria than the formal ones based on educational achievement (c.f. Breen and Jonsson 2007). As has been noted by Breen and Jonsson (2007), the relative homogeneity of the system of higher education, with small differences in prestige between institutions, may also contribute to the explanation. However, it remains unclear how these results and hypotheses might be translated into an understanding of the impact of parental mobility in relation to downward mobility. The safest conclusion would be to leave it open for future research to examine whether the impact of parental levels of education on the risk for downward mobility may be as powerful in other countries as this study has shown them to be in Sweden.


1As can be seen from the above quote, Richardson (1977) uses the somewhat value-loaded term “illegitimate mobility” to describe this, whereas occupational mobility in relation to levels of education is termed “legitimate”.

2Although many would agree that the work of Richardson has not received the attention it deserves, it is also true that a number decades have now passed since this work was conducted, decades in which society has changed in several ways. However, we do not believe that this affects the appropriateness of testing the hypotheses on data that were collected more recently.

3Richardson’s sample comprised 884 (both upwardly and downwardly) mobile individuals of whom a sub-sample of 117 were also interviewed. Of the 884 in the sample, 100 were downwardly mobile. Unfortunately it is not clear from the article how many of these were interviewed. However, in addition to the 117 individuals interviewed, the researchers also interviewed 40 of their fathers.

4As we know, don Quixote wanted to act the part of a noble knight, but lived in a world where such chivalrous behaviour was outdated and therefore ended up for the most part simply being laughed at.

51 Raymond Boudon’s (1974) notion of the primary and secondary effects of social origin on educational inequality has attracted much interest among researchers in recent years (e.g. Goldthorpe 1996; Nash 2003, 2006; Barone 2006; Jackson et al. 2007; van der Werfhorst and Hofstede 2007; Gabay-Egozi et al. 2009; Erikson and Rudolphi 2009; Tieben 2009). Whereas primary effects are those accounted for by differences in early academic ability between children from different social origins (irrespective of how these are generated), secondary effects are the result of differences in preferences and choice, once ability is taken into account. In the tradition following Boudon, primary effects are seen as the result of socialization, childhood living conditions and probably to some extent genetics (e.g. Erikson and Rudolphi 2009), whereas secondary effects are understood as the result of rational decisions made on the basis of cost-benefit calculations (e.g. Nash 2003, 2006). The mechanisms that generate primary effects are not thoroughly specified by Boudon or his followers however – their emphasis is instead directed at the rational choice based understanding of secondary effects. It has therefore been argued that Bourdieu has to date presented the only sociological theory whose principal aim is to explain primary effects, although opinions diverge as to whether he has succeeded (e.g. Lamont and Lareau 1988; Nash 2003; Barone 2006).

61 A prerequisite for a probability match is that the data sets have overlapping information, preferably from the same sources. This was the case for these two data sets. The probability-matching process involved distinguishing unique combinations of 13 variables included in both data sets for the individuals included in the Metropolitan study. (For a more detailed description, see Stenberg et al. 2006). The researchers who took the initiative and who are responsible for the SBC database are Denny Vågerö of the Centre for Health Equity Studies (CHESS) and Sten-Åke Stenberg of the Swedish Institute for Social Research (SOFI).

71 An even more narrow classification of parental occupational position would of course have made the studied group of subjects even more homogenous, but the limited size of the database means that we have to allow some variation in this variable.

8Codes 14, 21, 22.

9SEI-codes 11, 12, 21, 22, 33, 34, 35.

10SEI-codes 44, 45, 54, 55, 57, 60.


12As has been noted by e.g. Lareau (1987) there may also be changes over time in the kind of cultural capital that is most strongly associated with academic success, for example in relation to curricular change.

13It cannot be ruled out that the number of books in a household is to some extent also a function of economic capital.

14Although here the question relates to the number of shelf-metres of books, rather than the number of books.

15The principle adopted was to produce categories that were as similar in size as possible.


AHRNE, G., Roman C. & Franzén, M. (1996). Det sociala landskapet. Göteborg: Korpen.

ALM, S. (2001). The Resurgence of Mass Unemployment. Studies on social consequences of joblessness in Sweden in the 1990s. Dissertation series no 53. Stockholm: Swedish Institute for Social Research, Stockholm University.

ALM, S. (2008). Social nedåtrörlighet mellan generationer – en prövning av tre perspektiv. Stockholm: Institutet för Framtidsstudier, arbetsrapportserien.

BARONE, C. (2006). Cultural Capital, Ambition and the Explanation of Inequalities in Learning Outcomes: A Comparative Analysis. Sociology, 40 (6): 1039-58. [doi:://]

BECKER, R. & Hecken, A. E. (2009). 'Higher Education or Vocational Training? An Empirical test of the Rational Action Model of Educational Choices Suggested by Breen and Goldthorpe and Esser.' Acta Sociologica, 52 (1): 25-45. [doi:://]

BOUDON, R. (1974). Education, Opportunity and Social Inequality. New York: Wiley.

BOURDIEU, P. & Passeron, J. C. (1977). Reproduction in Education, Society and Culture. London: Sage.

BOURDIEU, P. (1984). Distinction. A social critique of the judgement of taste. Cambridge, Mass.: Harvard University press.

BOURDIEU, P. (1986). The forms of capital, in Richardson, J. G. (ed.). Handbook of Theory and Research for the Sociology of Education. New York: Greenwood Press.

BREEN, R. & J. Goldthorpe, J. H. (1997). 'Explaining Educational Differentials: Towards a Formal Rational Action Theory'. Rationality and Society, 9 (3): 275-305. [doi:://]

BREEN, R. (2001). A Rational Choice Model of Educational Inequality, Juan March Institute Working paper 2001/166.

BREEN, R. and Luijkx, R. (2004). Conclusion, in Breen, R. (ed.), Social Mobility in Europe. Oxford: Oxford University Press. [doi:://]

BREEN, R. & Jonsson, J. O. (2007). Explaining Change in Social Fluidity: Educational Equalization and Educational Expansion in Twentieth-Century Sweden, American Journal of Sociology, 112 (6): 1775-1810. [doi:://]

BROWN, D. K. (1995). Degrees of Control. A Sociology of Educational Expansion and Occupational Credentialism. New York: Teachers College Press.

CROOK, C. (1997). Cultural Practices and Socioeconomic Achievement. Westport: Greenwood Press.

DEARY, I. J., Taylor, M. D., Hart, C. L, Wilson, V., Smith, G. D. Blane, D. & Starr, J. M. (2005). Intergenerational Social Mobility and Mid-Life Status Attainment: Influences of Childhood Intelligence, Childhood Social Factors and Education. Intelligence, 33 (5): 455-72. [doi:://]

ERIKSON, R. (1984). Social Class of Men, Women and Families, Sociology, 18, 500-14. [doi:://]

ERIKSON, R. & Goldthorpe, J .H. (1992). The Constant Flux. Oxford: Clarendon Press.

ERIKSON, R. & Rudolphi, F. (2009). Change in Social Selection to Upper Secondary School- Primary and Secondary Effects in Sweden. European Sociological Review Advance Access published May 15, 2009.

FRITZELL, J. & Lundberg, O. (eds) (1994). Vardagens villkor. Stockholm: Brombergs.

GABAY-EGOZI, L., Shavit, Y. & Yaish, M. (2009). Curricular Choice: A test of a Rational Choice Model of Education, European Sociological Review Advance Access published June 5, 2009.

GOLDTHORPE, J. H. 1996 'Class Analysis and the Re-Orientation of Class Theory: The Case of Persisting Differentials in Educational Attainment', British Journal of Sociology, 47(3): 481-505. [doi:://]

GOLDTHORPE, J. & Jackson, M. (2007). Intergenerational Class Mobility in Contemporary Britain: Political Concerns and Empirical Findings. British Journal of Sociology, 58 (4): 525-46. [doi:://]

DE GRAAF, N. D., de Graaf, P. M., & Kraaykamp, G. (2000). Parental Cultural Capital and Educational Attainment in the Netherlands: A Refinement of the Cultural Capital Perspective, Sociology of Education, 73, 92-111. [doi:://]

HEMMINGSSON, T., Lundberg, I. & Diderichsen, F. (1999). The Roles of Social Class of Origin, Achieved Social Class and Intergenerational Social Mobility in Explaining Social-Class Inequalities in Alcoholism among Young Men. Social Science and Medicine, 49 (8): 1051-59. [doi:://]

JACKSON, B. & Marsden, P. (1962). Education and the Working Class: Some General Themes Raised by the Study of 88 Working Class Children in a Northern Industrial City. London: Routledge & Kegan Paul.

JACKSON, M., Erikson, R. E., Goldthorpe, J. H. & Yaish, M. 2007. Primary and Secondary Effects in Class Differentials in Educational Attainment: The Transition to A-level Courses in England and Wales, Acta Sociologica, 50, 211-29. [doi:://]

JANSSON, C-G. (1995). On Project Metropolitan and the Longitudinal Perspective. Project Metropolitan, Research Report no. 4. Stockholm: Department of Sociology, University of Stockholm.

JONSSON, J. O. (1987). Class Origin, Cultural Origin and Educational Attainment, European Sociological Review, 3, 229-42.

LAMONT, M. & Lareau, A. (1988). Cultural Capital: Allusions, Gaps and Glissandos in Recent Theoretical Developments, Sociological Theory, 6, 11-34. [doi:://]

LAREAU, A. (1987). Social Class Differences in Family-School Relationships: The Importance of Cultural Capital. Sociology of Education, 60, 73-85. [doi:://]

DI MAGGIO, P. (1982). Cultural Capital and School Success, American Sociological Review, 47, 189-201.

MILLER, M. (1960). Comparative Social Mobility, Current Sociology, 9 (1): 1-80. [doi:://]

NASH, R. (2003). Inequality / Difference in Education: Is a Real Explanation of Primary and Secondary Effects Possible?, British Journal of Sociology, 54 (4): 433-51. [doi:://]

NASH, R. (2006). Controlling for "Ability": A Conceptual and Empirical Study of Primary and Secondary Effects, British Journal of Sociology of Education, 27 (2): 157-72. [doi:://]

PARKIN, F. (1971). Class Inequality and Political Order: Social Stratification in Capitalist and Communist Societies. St Albans: Paladin.

VAN DER PLOUGH, S. (1994). 'Educational Expansion and Returns on Credentials', European Sociological Review, 10: 63-78.

RICHARDSON, C. J. (1977). The Problem of Downward Mobility. British Journal of Sociology, 28 (3): 303-320. [doi:://]

RODGERS, B. & Mann, S. L. (1993). 'Re-thinking the Analysis of Intergenerational Social Mobility: A Comment on John W. Fox's Social Class, Mental Illness, and Social Mobility.' Journal of Health and Social Behavior, 34 (2): 165-72. [doi:://]

SCB (1982). 'Socioekonomisk Indelning (SEI).' Meddelanden i samordningsfrågor, 1982:4, Stockholm: Statistics Sweden.

SOFI (2005). Project Metropolitan, Codebook I. Stockholm: Swedish Institute for Social Research.

SELLERS, S. L. (2000). Dreams Delivered, Dreams Deferred: Mental and Physical Health Consequences of Social Mobility among Black American Men and Women. (Doctoral dissertation). Ann Arbor: University of Michigan.

STENBERG, S-Å & Vågerö, D. (2005). Cohort Profile: The Stockholm Birth Cohort of 1953, International Journal of Epidemiology, 35 (3): 546-548.

STENBERG, S-Å, Vågerö, D., Österman, R., Arvidsson, E., von Otter, C. & Jansson, C-G (2006). Stockholm Birth Cohort Study 1953-2003: A New Tool for Life-Course Studies. Scandinavian Journal of Public Health, 35 (1): 104-10.

STOCKÉ, V. (2007). Explaining Educational Decision and Effects of Families' Social Class Position: An Empirical Test of the Breen-Goldthorpe Model of Educational Attainment. European Sociological Review, 23, 505-19.

SULLIVAN, A. (2001). Cultural Capital and Educational Attainment, Sociology, 35 (4): 893-912. [doi:://]

TIEBEN, N. (2009). 'Parental Resources and Relative Risk Aversion in Intra-Secondary Transitions: A trend Analysis of Non-standard Educational Decision Situations in the Netherlands.' European Sociological Review Advance Access Published November 23, 2009.

TIMMS, D. (1998). Gender, Social Mobility and Psychiatric Diagnoses. Social Science and Medicine, 9, 1235-47. [doi:://]

TOLSMA, J., de Graaf, N. D. & Quillian, L. (2009). 'Does Intergenerational Social Mobility Affect Antagonistic Attitudes Towards Ethnic Minorities?', British Journal of Sociology, 60 (2): 257-77. [doi:://]

VAN DER WERFHORST, H. G. & Andersen, R. (2005). Social Background, Credential Inflation and Educational Strategies, Acta Sociologica, 48 (4): 321-340. [doi:://]

VAN DER WERFHORST, H. G. and Hofstede, S. 2007 'Cultural Capital or Relative Risk Aversion? Two Mechanisms for Educational Inequality Compared.' British Journal of Sociology, 58 (3): 391-415. [doi:://]

WEST, P. (1991). Rethinking the Health Selection Explanation for Health Inequalities. Social Science and Medicine, 32, 373-84. [doi:://]

UniS: University of Surrey logo University of Stirling logo British Sociological Association logo Sage Publications logo Electronic Libraries Programme logo Epress logo