Most societies have a socioeconomically stratified system. A stratified system is defined as a "hierarchical ordering of positions in terms of power, privilege and prestige" (Kohn & Slomczynski 1990, p.31). Studies of social stratification use levels of occupational status to determine the locations of individuals in the hierarchy of the stratified system. The higher levels of status are assigned to more complex jobs and the lower levels to less complex jobs. Within the stratified system an occupation is assigned to an individual on the basis of her or his schooling and skills. If everybody has the same chances to achieve any occupational title that matches his or her education and skills, we call this equality of opportunity. Inequality of opportunity is the result of exclusion of individuals not by their personal ability, but by the enforcement of some form of ascriptive criteria.

By being born into a social stratum, thus through social origin, people sometimes receive status benefits before they enter the attainment process and prove their abilities. Here children 'inherit' the socioeconomic status of their parents. The most extreme form of applying ascriptive criteria is primogeniture, the right of the first-born to inherit the firm, a farm, title or rank. For this case we can say that social origin fully determines a child's status attainment. Usually, though, we encounter more subtle forms of ascription. Very often the jobs of parents and children are more or less similar or merely related. Research in social stratification studies this relationship between social origin and children's status attainment.

Besides social origin, the education of a child also heavily influences her or his job status later in life. The earliest study to model the exact importance of education for children's status attainment was carried out by Blau and Duncan (1967). They were the first to partition the pathways of status attainment into the dimension of 'achievement' and 'ascription'. Any status attainment of children that can be traced back to their own education is their personal achievement. On the other hand, any status attainment of children that can be traced back to their parents' education and occupational background is ascription of socioeconomic status (see Figure 1.1).

The classical status attainment model, developed by Blau and Duncan in 1967, captures the causal relationships between the education and occupational status of two generations: the father and the son. Their model is the point of departure for this current study (see Figure 1.2). Blau and Duncan's status attainment model contains five measures of socioeconomic status, two for the father and three for the son, ordered from the left to the right according to their occurrence in the life cycle: father's education and occupation precede the son's education. The father's education and occupation and the son's education precede the son's first occupation, and current occupation in 1962.¹ The father's education and occupation are so-called exogenous variables. These two exogenous variables influence the three subsequently occurring career steps of his son's status attainment: his education, first occupation after leaving school, and current occupation in 1962. These latter three career steps are the so-called endogenous variables in the model.

The most important feature to notice in Figure 1.2 is the extent of the relationship between the exogenous and endogenous variables, given by the numbers above the arrows, the coefficients. The size of the relationship between these five variables shows to what extent advantages are transmitted from one generation to the next. If the value of the coefficient between the exogenous and endogenous variables is high, the transmission of advantages is high and socioeconomic mobility is low. Vice-versa, if the value of these coefficients is low, the transmission of advantages is low and socioeconomic mobility is high.

Thus, Blau and Duncan rephrased socioeconomic mobility by measuring the influence of the father's education and occupation on the son's education and his first and present occupational status. Their model provides us with a tool to dissect the relationship between social origin of the individual and his or her attained occupational status. The elliptic line on the left-hand side indicates the correlation (relationship) between the two exogenous variables that is not analysed. A direct influence, here called a path coefficient, is drawn as a straight line. The entire path model partitions the correlations between all variables into direct, indirect and spurious effects. Remember from Figure 1.1 that the path running between the education and occupational status is an achievement relationship and the path between the socioeconomic background of the parent and the child is what we call ascription by social origin. The coefficients in Figure 1.2 tell us that the direct or net effect of the son's education on his first occupational status is about twice as high (0.440) as the direct or net effect of the father's occupation on the son's first occupation (0.224). This ratio increases later in their careers. We observe that the direct effect of the son's education on his current occupation in 1962 was about three times higher (0.394) than the direct effect of the father's occupation on the son's current occupation in 1962 (0.115).

If we compare the total effects of the son's education and the father's occupation on the current occupation of the son, this ratio is higher. Part of the influence of the son's education and the father's occupation is transferred via the first occupation of the son. Total effects can be calculated by multiplying the effects of the son's education and father's occupation by the first occupation of the son. The total effect of the son's education on his current occupation in 1962 is 0.518 (=0.394+0.440*0.281), whereas the total effect of the father's occupation on the son's current occupation in 1962 is 0.178 (=0.115+0.225*0.281). Thus, the total effect of the son's education is three times higher than the total effect of the father's occupation. Clearly, in 1962 achievement was more important than ascription.

It was a considerable accomplishment of Blau and Duncan to measure the social mobility between two generations and show how the father's effects develop over the life cycle of the son in a single model. For the first time status 'ascription' by parents and children's individual 'achievement' were accurately dissected in the process of 'status attainment'. Empirically they implemented their model by estimating the influence of the father's status transfer on the son's education and occupation from a large, nationally representative sample of men in the USA in 1962. For that period - during the 1960's - the empirical calculations they carried out were remarkable because the computing facilities were still basic. Now their procedure has become standard in parts of sociological research.

Yet, men form only half of the population and they always have, apart from a father, a mother as well. From the start, the OCG study was not designed to include daughters and only one of the tables in the research report of Blau and Duncan includes some information on the influence of the mother's education on the son's education. The result in this table suggests that this effect is as large as the father's (p.189). However, in a report containing more than 500 pages this remarkable result is mentioned, to my knowledge, only once: "The net, or direct, effects of these characteristics of the wife, though they are modest in magnitude, cannot be dismissed as chance findings" (p.345). However, Blau and Duncan then go on to dismiss this result on other grounds. They assumed that the mother's effects "[...] would disappear in a system of variables including one or more strategic characteristics of the husband that we failed to measure" (ibid.).

Historians of stratification research have been quick to point out that Blau and Duncan (1967) are in excellent company with other renowned researchers of social mobility when it comes to excluding the influence of the mother from the scope of the research (Ganzeboom et al. 1991). In the first generation of social mobility studies (e.g. Glass (1954) for England and Wales and Van Tulder (1962) for the Netherlands) women's mobility did not appear in the research reports. In the second generation of social mobility studies, which were often a replication of Blau and Duncan's study, mothers and daughters were an either omitted or were an under-represented group compared to fathers and sons (Featherman & Hauser 1978, Goldthorpe et al. 1972).

Several rationales exist for the exclusion of mothers and daughters from the scope of status attainment research. The three most prominent justifications are: first, if they ever enter paid employment, women commonly stop working as soon as they marry or have children. Secondly, even if mothers remain employed, their occupational status hardly has any influence on the status attainment of her children, because they, on average, have a lower educational level and occupational status than fathers and thus fewer socioeconomic resources to transfer. A third objection to the inclusion of mothers is less based on a substantive reason. Many times researchers simply find it too difficult to include the socioeconomic status of mothers in their study. Because of their intermittent labour market participation, researchers encounter missing information on the occupational status of mothers and daughters. The following section will offer some arguments for why these assumptions may no longer hold. It shows the development of female educational attainment, economic activity and occupational status over recent decennia.

The following description of women's educational level, labour market participation and occupational status focuses on the developments found in the USA, the Netherlands and Germany. This choice of countries was made, because in the further course of the study much of the analysis will be based on either one or more of these countries. However, many of the developments described here are not unique to this current selection but can be found in most Western industrialized countries. The observation window spans the last three to four decennia. A longer observation window would have been preferable, but internationally comparable data are difficult to acquire for earlier years. In the following section, the figures shown are based on the adult population and if possible, restricted to married women and men, and thus include much of the target population for the empirical studies later on.²

The educational status of mothers is often assumed to be lower than the educational status of fathers, and because of this they are often excluded in research on social stratification. In the following two sections we will look at the development of the educational level of women over the last three to four decennia. Subsequently we will study the differences in educational level between husbands and wives. The surveys used to obtain these figures are part of the International Stratification and Mobility File (ISMF, Ganzeboom & Treiman 1999).³ The numbers shown refer to the year when the survey was held.

First we compare how the average number of years in education of men and women has developed in the USA, Germany and the Netherlands (see Figures 1.3.a to 1.3.c). The graphs show the average number of years spent in education for the various years when the survey was held (see appendix A). The dots in the graph refer to the raw measurements from population samples. An estimated linear trend indicates the development of the average educational level of men and women in each of these countries. Note that for Germany no older data than a survey from 1969 was available, therefore we have no information for earlier years. For men and women we notice an upward trend in the number of years of schooling.

In the USA the average number of years women spend in education is higher than in the Netherlands or Germany. In the USA, women have, compared to men, almost the same average level of education. Over time the average number of years in education have risen more slowly in the USA than in Germany or the Netherlands. There, the estimated trends suggest that within roughly 35 years of observation the average duration of education for men and women has increased by two years. In Germany the average duration of education has increased by two years between 1968 and 1988, rising from a little over eight years in 1968 to approximately 10 years in 1988. The Netherlands show the most dramatic development of trends in duration of education. Here the average duration of education has risen from almost eight years in 1958 to almost 12 years for men and 11 years for women of years spent in education in 1996.

In Germany the upward trend for women is slightly higher than for men. Although in Germany women started further down the scale than men, they are catching up with the average duration of men's education. At the end of the observation window women still lag slightly behind men, but for Germany the gap is closing. For the Netherlands and the USA the trend towards a closing gap between men's and women's education is not as pronounced. However, one has to consider that trend lines on the aggregate include observations at all age levels. If we compare trends on the average duration of education between national populations the development is not as pronounced as if we undertake the same comparison between age groups.

We have seen that on the aggregate level the duration of education of women in the USA, Germany and the Netherlands is a slightly lower than that of men. However, these trends do not automatically imply that a woman with a lower level of education is married to a man with a higher level of education. The following section will look at this latter argument more closely.

The next comparison again is based on the ISMF (International Stratification and Mobility File, Ganzeboom & Treiman 1999), but this time for the differences in educational level between husbands and wives, in the USA, Germany and the Netherlands. The difference scores are displayed in Figure 1.4.a to 1.4.c. The black shaded area shows the percentage of couples where the husband's education exceeds his wife's education by (at least) two years. The white area indicates the percentage of couples where the husband's and the wife's education are no more than two years apart. These couples were coded to be 'status equal'. The grey shaded area displays the percentage of couples where the wife exceeds her husband's education by at least two years. Note that the observed percentages have been smoothed out by estimating a linear trend.In the USA over the years there has been a trend towards more equality between husbands' and the wives' educational levels (Figure 1.4.a). At the beginning of the 1960's the percentage of marriages where the duration of the husband's education exceeds his wife's is about 30%. But wives also exceed husbands' educational level in 30% of the marriages. This leaves 40% of marriages where wives and husbands are status equal. In 1990, 20% of marriages consist of a husband with a higher education than his wife. Almost the same is true for the reverse case; the percentage of marriages where the wife has a higher educational level than her husband is roughly 20% as well. In 60% of the marriages educational status equality exists between spouses.

Figure 1.4.b shows the percentage of marriages with unequal and equal education for husbands and wives in Germany. The percentage of marriages where wives are better educated than their husbands increases over the years. In 1969 about 10%, whereas in 1992 roughly 20% were marriages where wives were better educated than their husbands.

Educational status equality between spouses is largest for Germany, compared to the USA or the Netherlands. Between 1969 and 1992 the percentage of marriages where the husband exceeds his wife's education duration has remained stable at approximately 65%. Figure 1.4.c shows the development for the Netherlands. The percentage of marriages where both spouses are equal status increases between 1970 and 1996. However, in 1970 in more than 20% of all marriages and in 1996 in fewer than 20% of all marriages, the wife's education was higher than that of her husband. The percentage of marriages in which the husband exceeds his wife's education remain more or less stable at 40% between 1970 and 1996. The case that the husband exceeds his wife's educational level is less common than the cases taken together where they either have an equal educational level or where the wife exceeds her husband's educational level. Therefore, concerning the education of husbands and wives, the case that the husband exceeds his wife's education has been overstated.

For all three countries and throughout the observation window, a trend towards a continuously increasing labour market participation of women can be observed (see Figure 1.5). When comparing the rate of female labour market participation for the Netherlands, Germany and the USA, we see that the Dutch rate used to lag far behind that of the other two countries. Since mid 1980, however, it shows the steepest increase.

The female employment rate in the USA has always been higher than for Germany or the Netherlands. German women have occupied an intermediate position between the Netherlands and the USA. Their employment rate also shows an upward dynamic, but it has been slower compared with that found for the USA and the Netherlands. In 1989 the reunification of East and West Germany took place. Labour market participation of East German women used to be much higher than that of West German women. Therefore, after reunification, the number of employed women slightly increased.

Of course, the next question is whether the overall trend towards an increased rate of employed women applies to all age groups in a similar way and whether it holds also for women who are mothers. Perhaps only part of the female population, i.e. young single women, are responsible for the development. Perhaps it is still pertinent that as soon as women have family obligations they leave the labour market in large numbers to care for their children and family.

Figure 1.6.a to 1.6.c show the active female population⁴, 16 to 75 years old, in the USA, Germany and the Netherlands. For all three countries in 1979, women aged between 20 and 24 years have the highest rate of economic activity. It decreases for women who are between 25 and 34 years old. This decrease is most pronounced in the Netherlands.

In 1979 many women never re-enter the labour market. The economic activity rate for women between 45 and 54 years in the USA is around 60%, for Germany it is around 50%, and for the Netherlands it is at the 25% level. To some extent it is thus true that women used to quit employment when they reach the age to marry or to have family obligations. Many chose to be homemakers.

However, by 1988 we observe that the distribution of the active female population in the USA shows a small 'dip' for women between 25 and 39 years of age. It indicates that a certain proportion of mothers, when their children are young, stay at home and care for the children. Later in life, when they are around 40, they often become economically active again.

The lowest labour market participation we notice for women between 30-34 years. Neither for Germany (West) nor for the Netherlands do we see the same pattern as for the USA. By 1987 the Netherlands shows a similar distribution of economically active women to that in Germany; in both countries only up to the age of 24 are the majority of women employed. In 1987, women in East Germany had a far higher rate of female employment than women in the USA, West Germany or the Netherlands. This is because in the German Democratic Republic the government enforced a high rate of female employment.

In all three countries we notice a peculiar development during the most recently observed year of 1997. The age group that shows the highest rate of economically active women shifts from the age between 20 and 25 years to the age of 25 to 29 years. The explanation is that women spend an increasing amount of time in education and postpone childbearing. Not the Netherlands but the other two countries have female participation rates that are becoming more bell-shaped. All three figures indicate that at one time or another more than 60% of the female population have been economically active in these countries. Recently, to an increasing degree women's economic activity tops at a stage where family and child rearing obligations are liable to occur, that is between the ages of 25 and 45. We can conclude that many women, even when they have become mothers, continue to work. Discharging mothers from the research agenda on the basis of their economic inactivity is no longer a valid argument.

The next argument for excluding the influence of mothers from studies in social inequality is that if wives are employed they usually have an occupational status lower than that of their husband, and therefore fewer resources to transfer to their children. For the USA, Germany and the Netherlands I show in Figures 1.7.a to 1.7.c how the occupational status scores between husbands and wives have differed over the last three to four decades. The data points were smoothed out, that is a linear trend was estimated, in order to facilitate their interpretation. The occupational scores were computed, based on the ISEI index of occupational status (Ganzeboom et al. 1992) and range between 10 and 90 (the nature of the ISEI will be discussed later on in section 1.6).The division of marital status differences is threefold. In the first group the husband exceeded his wife's occupational status by more than eight points. The black bars relate to the percentage of couples in the first group. In the second group, from here on called 'equal' status, the differences between the husband and the wife were no more than eight points. The white bars indicate the percentages of marriages in the second group. In the third group the wife exceeded the husband's occupational status by more than eight points. The grey bars show the percentage of marriages in the third group. In the USA (Figure 1.7.a) we observe a trends towards occupational status dominance of the husband.

In the past there were a higher percentage of marriages where wives' job statuses exceeded husbands' job statuses (roughly 40%) than for the case where husbands' statuses exceeded wives status (around 20%). Within the last decade there are almost as many marriages where the husband is status dominant as there are where the wife is status dominant (both roughly 25%).

In contrast to the USA, Germany has a higher percentage of marriages where the husband exceeds his wife's occupational status (29.1%) and a lower percentage where the wife exceeds her husband's job status (27.5%). Still, in Figure 1.7.b we also notice a high percentage of marriages where the husband and the wife have an almost identical occupational status. We observe a slight trend towards a higher percentage of marriages in which the wife holds a higher occupational status than her husband, comparing the grey shaded area of the earliest survey (1969) with the latest survey (1992).

In the Netherlands (Figure 1.7.c), compared with the USA and Germany, we find the highest percentage of marriages where the husband's job status exceeds that of his wife by more than eight points (average over all surveys: 39.3%) and the lowest percentage for the reverse case, that the wife exceeds her husband's status (average over all surveys: 24%).

A trend towards less status equality exists. However, for a higher percentage of marriages the husband's occupational status is higher than the wife's occupational status compared with the reverse case, where the wife's occupational status is higher than the husband's job status. In recent years this has become increasingly the case.

In all three countries we notice that in more than 50% of the marriages either the wife exceeds the husband's job status or both spouses have an almost equally high occupational status. Taking all survey years together, on average of 50% of all married couples in the Netherlands, 51% in Germany and 67% in the USA have equal status spouses, or a wife who exceeds her husband's occupational status. Moreover, the percentage of marriages in which the husband's occupational status exceeds the wife's status is only marginally larger than the percentage of marriages where the wife exceeds her husband's status. For the USA this ratio is 32:31, for Germany it is 29:28. The Netherlands ranges outside this with a ratio of 39:24. Overall, therefore, the assumption that the husband usually has a higher occupational status than his wife has been also overstated. Excluding the influence of the mother on the basis of this assumption can no longer be viewed as a valid argument either.

Altogether the conclusions from the empirical evidence are: (A) Mothers will have on average almost the same educational and occupational level as fathers, in recent times increasingly so. (B) At some time in their lives almost all mothers will have held an occupational title of their own, so that even if they are currently out of the workforce, it is nevertheless possible to retrieve their occupational title from the time they had paid employment. (C) We can assume that spouses have equal status in at least half of all the cases.⁵

Of course, the argument that it is technically difficult to include mothers in research on social mobility also has to be considered (e.g. Ganzeboom et al. 1991, p.293). Yet, this should merely stimulate ideas about the solutions on how to overcome these problems. Although right up to the present day the influence of the mother has remained a largely neglected area in mainstream social mobility studies, some studies exist where interesting methods, models and questions have been proposed regarding the mother's influence. The following literature review may provide some good examples.

1.4 Literature Review

In contrast to what is commonly believed, studies on women's occupational mobility started to appear rather early (e.g. Hughes 1949, Ellis 1952, Caplow 1954). In line with the moral standards of those days, they dealt with psychological aspects of unmarried career women (Ellis 1952) or the "marginal man", i.e. the discrimination women faced in the labour market (Hughes 1949). Ellis (1952), for instance, compared upwardly mobile to non-upwardly mobile women. Her main hypothesis was that upward mobility was an outgrowth of basically neurotic drives resulting from unsatisfactory early primary group relationships. She found mobile women to be more socially isolated and maladjusted than non-mobile women. She did not fail to point out, however, that group differences were not as marked as her initial hypothesis had assumed them to be. Nevertheless, up to and including the 1950's, compared with the huge interest in male mobility processes, studies on females received minor attention. The interest in the role of the mother in social mobility starts to rise during the 1960's, but large scale empirical research on the subject was not to appear until the 1980's. We see that including the socioeconomic background of the mother provides insight into, for instance, the process of status formation (Vellekoop 1963), family ranking (Barth & Watson 1967), or drug and alcohol abuse of children (Haug 1973). Starting from 1970, some theoretical objections to traditional models of status attainment are raised. Research appears that opposes conventional assumptions (Acker 1973), postulates more extended models (Falk & Cosby 1975) and questions whether male-based results are applicable to the occupational outcomes of women (van Doorne-Huiskes 1984, Hörning 1984). Despite all these activities, Acker concludes in a literature review on women and stratification: until 1980 "[s]tratification theory has been a theory of white males" (1980, p.33). Let us now see whether, two decades later, her conclusion is still valid.

Investigations on how the mother's status background relates to her children's status attainment are up to this day dominated by the North-American literature. However, the pioneering study carried out in the USA to measure the importance of status transfer between generations, Occupational Changes in a Generation (OCG) of Blau and Duncan (1967), described in detail above, dealt only with sons and their fathers. After the accusation that the field of social stratification is 'a case of intellectual sexism' (Acker 1973), an increasing number of researchers started to include women in their studies on educational and occupational mobility.According to the classical status attainment model, the influence of family background on children's status attainment unfolds on three different levels. The educational attainment of children is influenced by the educational level and occupational status of parents. The occupational status of the child is influenced only by the occupational status of parents. Parental educational level has no influence on the occupational level of children (Blau & Duncan 1967, De Graaf & Luijkx 1992). The main body of research that focuses on the influence of the mother's status has chosen to study either the influence of the mother's education or her occupational level.

The extent of the influence of the mother's education on children's education varies from study to study, though overall she has been found to have a marked impact. In the OCG the effect of the mother's educational level on the education of her son was as large as that of the father. Many studies report that parents' educations affect children of both sexes, but that the mother has a stronger impact on her daughter than on her son (Treiman & Terrell 1975, Marini 1978, Peschar 1987, Miller & Hayes 1990, Crook 1995, Van der Lippe et al. 1995). Sewell et al. (1980) show that mothers are important only for daughters and do not affect their son's education at all. These results have led in one case even to an exclusion of sons from the empirical study (Hayes & Miller 1989). The latter study shows that the father's education is more important than mother's education for determining the education of daughters. Other studies suggest that the influence of both the mother's and father's educational background remain important for sons and for daughters. This has been found for the USA (McClendon 1976, Holland Baker 1989, Kalmijn 1994), Germany (Henz 1995) and some socialist countries (Peschar 1987, Hanley & McKeever 1996). For five socialist countries, Hanley and McKeever record an equal increase in the influence of both parents' education on children's education (1996). To summarize, the evidence supporting the same-sex role model is as large as the evidence rejecting it.

Trend analysis in the Netherlands suggests that historical changes regarding the influence of parental education on children's education are taking place. The mother's educational background, compared to the father's educational background, has gained influence between 1950 and 1980 (Bakker & Cremers 1994, Van der Lippe et al. 1995), partly compensating for the diminishing effect of the father's background during that time. An assumption that may explain this fact is that mothers are gaining power in family relationships because of their increasing economic independence and because of their increasing level of education.

If the mother's influence is growing because of her increased economic independence, then the impact of her occupational status on the educational attainment of her children should also increase. This relationship has indeed been established in the USA. According to Kalmijn (1994), the relative influence of the mother's occupational status compared with the father's on the education of children has increased over time. No significant trend over time for the influence of the mother's job on children's education was found for socialist countries (Peschar 1987, Hanley & McKeever 1996).

However, empirical work regarding the influence of the mother's occupational status on her children's educational level has been more scarce than on her educational status. For the Netherlands, Dronkers (1992, 1995) demonstrates that the mother's occupational level affects both her son's and her daughter's educational attainment positively and that working mothers have better educated children than homemakers. This study shows, however, that there may be one exception. If the mother has a blue-collar occupation, then this affects her child's educational attainment more negatively than if she is a homemaker. For the latter case, Dronkers concludes, the child is better off if the mother is a homemaker.

Studies that measure parents' occupational status on a continuous prestige or status scale find a significantly positive relationship between children's education and mother's and father's job status (Treiman & Terrell 1975, Holland Baker 1984, Hayes & Miller 1989, Miller & Hayes 1990, Crook 1995). Sometimes sex-role patterns are found, sometimes not. Whereas in Treiman and Terrell's (1975) early study, by the mother's job only her daughter's but not her son's education is affected; other studies find no sex differences (Hayes & Miller 1989, Miller & Hayes 1990, Crook 1995). Almost all above cited studies imply that the effect of the occupational status of the mother on the education of children is about half that of the effect of father's occupation (see for an exception: Holland Baker 1984).

At this point it is safe to conclude that we can expect the mother's socioeconomic background to profoundly and significantly influence the educational attainment of her children independent of the father. This is true for mother's educational as well as for her occupational status. However, with regard to the influence of her occupational level on the education of her children, the literature leads us to expect that the influence of the mother is less than that of the father.

Two different methods are used in the literature to determine the occupational influence of the mother on children's occupational choice. The main body of research uses bivariate mobility tables; that is, the occupations of mothers are grouped into six to eight different types, e.g. professional, managerial or clerical, etc. Subsequently the diagonal cases, representing inheritance, are compared with the off-diagonal cases. Often this is done only for mother-daughter dyads (Pearson 1983, Hayes 1987, Hayes 1990). If the influence of the father is excluded, strong inheritance effects are found between the mother's occupational class and her daughter's job destination. If the father's occupational class is included, studies show that his job is very important also but that the mother's job remains a strong predictor of the daughter's occupational destination (Rosenfeld 1978, Aschaffenburg 1994, Khazzoom 1997).Aschaffenburg (1994) points out that the status inheritance between blue-collar mothers and daughters is greatest. Mothers working in professional or managerial positions set more of an example for their sons than for their daughters. On the other hand, mothers who have entered into non-traditional (i.e. less female-typed) occupations are more highly related to their daughter's occupation than mothers working in traditional occupations. She concludes that the mother's occupational status is important both for sons and daughters, but that the reasons why differ between the sexes. This conclusion is shortly thereafter challenged by Khazzoom (1997). In contrast to Aschaffenburg's premises (1995), Khazzoom reports that if the mother is working in a professional position, the daughter is most likely to work in such an occupation too. This effect is doubled if the father is also working in a professional position. Interestingly, Khazzoom also shows that the decrease of the father's background influence, usually reported in over-time comparisons, diminishes if the mother's occupational background is included in the analysis.

The second method used is to place the mother's influence into the classical model of status attainment (Blau & Duncan 1967). Mobility tables, although showing the total inheritance effects between generations, neglect to control for the effects of one's own individual achievements, i.e. the children's own educational level. Here evidence becomes scarce. As the present study works along this paradigm, results of studies using the classical model of status attainment are very relevant. Treiman and Terrell (1975) estimated the net effects of the mother's job status on daughters' occupations and found a significantly positive relationship. Other studies have replicated this finding (Hayes & Miller 1989, Crook 1995). Henz shows that the mother's occupational background was important only for women in an earlier period, born in the 1930's (1996). She concludes that for younger cohorts and for sons particularly, the mother's occupational status has no direct impact. On the other hand, a Canadian study by Steven and Boyd (1980) goes so far as to suggest that the knowledge of the father's occupation is superfluous when predicting the daughter's occupational destination. Among all other results produced, their conclusion can be regarded as an exceptional one.⁶ Despite these contradictions in almost every study, we see that mother's job status is less important for predicting the occupational outcomes of sons as compared with daughters (Holland Baker 1983, Stevens & Boyd 1980, Henz 1997, Khazzoom 1997, but for the exception: Aschaffenburg 1994).

Most of the above studies suffer from severe limitations, though. For example, Treiman and Terrell (1975) do not control for the influence of the father's occupational status on the daughter's occupational status. Holland Baker (1984) does not use a representative sample to investigate the effects of parental background. Her data is limited to a small sample of mothers who gave birth to a child in 1948 or 1949 in one 'typical Midwestern city' (p.239). Hayes and Miller (1989) limit their study to daughters only. Henz's (1995) study, although using a representative sample, suffers from a small sample size.

None of the studies mentioned above use the first occupational status of children, after they have finished their formal schooling, to study the influence of the mother. Only the child's present occupation at the time of the interview is considered. This point may seem negligible, as former and later occupational status are closely related. However, in a study on the influence of the mother's status, looking only at current jobs has profound disadvantages. The most prominent disadvantage is that the influence of parents is likely to taper off as the occupational career of the child continues. The second disadvantage is that many daughters possibly have intermittent occupational careers because of family obligations. Therefore the results regarding the influence of the parents on sons' and daughters' job status later in life may suffer from serious bias, as their attained job status later in life, strictly speaking, may not be comparable.

Results so far have not produced an unequivocal picture. Cross-national comparisons on the influence of maternal educational and occupational level on the occupational status of sons and daughters have, to my knowledge, not yet appeared. Furthermore, models and measures of the influence of maternal background vary greatly between the studies and complicate the assessment of their outcomes.

The discussion on how to include women in studies on social inequality has been handled in two essentially disconnected manners. In the USA the focus was on the size of maternal status transfers on the children's educational and occupational attainment. Within the British research tradition, theoretical implications and empirical findings have focused on whether, and if so, how, to include wives into the study of the class position of families. The following debate is interesting, because it reveals some of the conservative attitudes that prevailed up until 1980 in the research community which occupied itself with studying intergenerational occupational mobility. However, this debate must not be viewed from merely an observational point. At a later stage in the British debate some new theoretical ideas were developed which tried to solve the problem of how the measurement of parental background could be optimized. These British models are quite useful for the current study while placing the mother and her influence at the centre of this inquiry into social inequality, in addition to the influence of the father. Some years passed, though, until the discussion had reached this point of departure. One of the most cited studies of British research on class formation in the early 1950' completely lacks any reference to women. In a nationwide study of Britain in 1949, Glass (1954) used only the status of the father or the husband to study class relationships. Information on the status background of wives was available from the questionnaires, but ignored. Later they let this information be destroyed (as is a routine procedure in civil service practice) before carrying out any analysis. Nowhere do Glass et al. claim to have tested their implicit assumption why they ignore women's own class positions; that is, that women stop working after childbirth. However, this and the assumption that the distribution of women among jobs is different from that of men will later become widely disputed subjects: "In our society, [...] the occupations held by women tend to be of lower status than those which men of a comparable background and education would be willing to accept" (Glass 1954, p.178). Almost three decades pass until a full-fledged discussion started on these issues.

Initially, the main argument for the inclusion of wives is that their employment is disproportionally more important for households with an unskilled or manual head than otherwise, because wives' economic contribution is so much more important in these families (Garnsey 1978). Despite some outcries against employing "double standards" (Haug 1973, Delphi 1981), these predictors continued to be used. Slowly, however, the tide turned and researchers started to ask whether "women's jobs 'make a difference'" (Britten & Heath 1983, p.56). They conceded, that "[...] the classification of social class which takes women seriously is both easier and more effective than conventional wisdom has allowed" (ibid., p. 60). Two problems remain. The first issue is how to make occupational scales comparable for men and women. The second question is, how to deal with cross-classified families that have resulted from the inclusion of wives' class backgrounds into the analysis.

Before these problems were tackled, Goldthorpe (1983, 1984) launched his widely cited defence of the conventional view. He contends that the wife's contribution to her family's class position through her earned wage is minor, compared to her husband's, and that the wife's employment generally ranks below her husband's employment (Goldthorpe 1983, p.473ff.). By claiming that the member of the household with the highest status determines the market position of the family, he dismisses the issue of working wives as a minor problem for the conventional view of class analysis. His allegations were swiftly answered.

After a re-analysis of Goldthorpe's data, Stanworth (1984) concludes that the wives' subordinate classes are systemic, rather than negotiated within the family. What is more important, Heath and Britten (1984) undertake a first attempt to reclassify wives' occupations and single out their impact on fertility decisions and voting behaviour. Methodologically speaking, the problem is that women often work in clerical, non-manual white collar jobs, but as "lower grade" employees, an occupational background so far overlooked in occupational class typologies (e.g. Goldthorpe 1987). They show that "[...] the women's own qualifications are more important than their husband's class as an explanation of their career paths" (Heath & Britten 1984, p.486, emphasis by authors).

A modification of the conventional paradigm is the 'dominance' model (Erikson 1984). Erikson argues that if we relate the market positions of a family to the person with the highest class background, the husband's background is the correct basis of the analysis only if his wife holds an inferior class position. Following his suggestions, Goldthorpe and Payne (1986) concede that the mobility of women is "grossly impaired" if they apply the conventional view, because women then display downward mobility much more often than men (p.548f.).

The next issue tackled, is the problem of how to deal with cross-class families. It was initially studied using qualitative analyses (McRae 1986, Leiulfsrud & Woodward 1987, 1988). From these studies the impression emerges that cross-class families, where one spouse holds a class position diametrically opposed to the other, have a different sort of class behaviour, other cultural resources and power relationships different from homogeneous class marriages. At this point, the idea of a 'joint' classification emerges (Marshall et al. 1988). A few years later Graetz (1991) introduces an empirical strategy and Sørensen (1994) a theoretical model for the joint classification (for further details see Chapter 2).

Up to this point, it has been established that a child's status attainment is profoundly influenced by her or his mother's status background. As stated before, the primary goal of the current study is a systematic approach to the analysis of the influence of the mother's status background, in relation to that of the father's, on children's educational and occupational status attainment. In this respect it will be of particular to see interest how the relationships in the classical status attainment model change if we add the mother's influence to that of the father. As an overriding research aim we can identify the problem of whether and how the mother influences her children's status attainment and now proceed to specify this term more closely. The first basic question is the extent of the mother's influence and whether or not she has an influence on the status attainment of her children at all. In the second instance, the influence of the mother is compared to the influence of the father, because he is known to be an important source of the transfer of status resources. The current study answers how the mother, in relation to the father, influences the status attainment of children. The third question is how far the mother's status background has a special impact on her daughter, as compared to her son's status attainment. The focus is on the importance of the sex-role model for intergenerational status transfer. The fourth question is how the mother's influence has changed over time, in relation to that of the father, because the increase of maternal status resources in recent times may have caused some changes in trends in intergenerational status transfer.

Chapter 2 and 3 of this book contain empirical studies on how the mother's education and occupational status influence children's educational attainment. In Chapter 2, the problem to be solved is how the influence of parental background can be modelled most efficiently. In the literature reviewed above we have seen that various concepts exist. These various concepts have not yet been put to an empirical test that would allow for a comparison of their explanatory power. Therefore, up front the analysis seeks to show which is the best model to measure the impact of social origin on status attainment, if, in addition to the father, we also study the influence of the mother. The requirements the empirical model has to meet are straightforward. Both parents' education and occupation should be considered, together with historical trends of the influence of social origin. The latter is decisive for a study of social inequality, if we want to be able to judge the development of how important the mother has been, compared to the father, on the process of stratification. Up until now it remains unknown how the influence of her occupation in addition to her educational level has developed throughout recent history.

Analyses so far have been restricted to including the influence of the mother's education only (Bakker & Cremers 1994, Van der Lippe et al. 1995, both for the Netherlands). Thus, the mother's impact is best documented for the educational attainment of her children. Studies that include both the mother's education and her occupational status have remained scarce and nationally restricted (Kalmijn 1994, Henz 1995). Although large scale international comparisons of the influence of the father on the educational attainment of children have appeared frequently in recent years (Treiman & Yip 1989, Shavit & Blossfeld 1993, Rijken 1999), a similar approach that includes the influence of the mother is still awaiting application (although see for socialist countries: Hanley & McKeever 1996). Based on the evidence found so far, we can expect a significant effect of the mother's education and occupation, independent of the father's socioeconomic background (Kalmijn 1994, Crook 1995). The resulting paths in the status attainment model are drawn in Figure 1.8. The main issue covered by the second chapter is how the mother, in addition to the father, influences children's educational attainment. The focus is on the following research questions:

1. What is the most appropriate model to study how mother's, in relation to father's socioeconomic status has influenced children's educational attainment over recent decades?
2. How do conclusions about less educational reproduction change if, in addition to the father, the influence of the mother's socioeconomic status is also considered?

In Chapter 3 a more theoretical research question will be answered. For several years now there has been a debate on the question of whether it is better for a mother to stay home and care for her children instead of taking up out-of-home employment. The scientific and public argument opposed to maternal out-of-home employment holds that the restricted time of employed mothers may have a negative impact on children's school attainment. On the other hand, status attainment research shows that a linearly positive relationship exists between mother's job status and the education of her children.

Because the mother is the main attender of the children, it may be the case that for mothers, other than for fathers, not only are their status resources important, but also their time restriction may influence the educational attainment of their children. Up until now it has remained unresolved just to what extent these two factors weigh for the educational attainment of her children. It may be the case that the socioeconomic resources of the occupational status that an employed mother has acquired compensate for many of the negative effects of labour market participation per se. Here the relationships as shown in Figure 1.9 are studied. Although the main issue is again the influence of the mother on the educational attainment of her children, the research question here is explanatory and reads: How heavily do the time restrictions, caused by the mother's employment, and her occupational resources influence children's educational attainment?

The focus in Chapters 4 and 5 is on mother's influence on the occupational status attainment of her children. Commonly studies that analyse the size of intergenerational occupational status transfer use the current, not the first, occupational status of children. Some disadvantages that are connected to this strategy have already been discussed above. One of the advantages of our approach is the high probability that adult children of either sex will have at least one entry job after they finish their school. As child rearing responsibilities are likely to occur later in life, the comparability between men and women's entry positions is high.

Indications exist that the same-sex role model may be important for the occupational attainment of children. In Chapter 4 an explicit empirical test will be carried out regarding this expectation. However, children may be inclined to follow their parents' example not only regarding their occupational status. When studying the influence of the mother's job on her children's jobs, considering the effect of occupational sex-typing may be important, too. Women seem to be much more disadvantaged by the sex-typing of their job than men. The mother's occupational sex-typing may form a second opportunity for status transfer that determines her children's occupational status attainment. Figure 1.9 shows which of the paths in the status attainment model are used in Chapter 4. The main issue in this chapter is how intergenerational transfer patterns of occupational status and sex-typing change, if we add the mother's background to the classical model of status attainment. The following research questions are answered:

1. How does the status and sex-typing of the mother's, in relation to the father's occupation influence the status and sex-typing of the occupations of daughters and sons?
2. How have these relationships changed over time?

If we want to observe how the influence of the parents develops during children's careers we have to consider the child's career dynamics. For instance, knowing when the child held what kind of status during her career is important, but knowing how her job transitions are influenced by her social origin provides additional insights. Although life history techniques have now been around for several decades, they have not yet been widely applied to study the influence of the mother and the father in a dynamic perspective.

Studies almost unanimously point to the fact the mother's occupation has a stronger impact on her daughter than on her son's occupational location. Some researchers have suggested that daughters remain closer to their mother's occupational location, than sons to their father's occupational location, as their careers advance (e.g. Dex 1987, 1990). Possibly the stronger orientation towards the example set by the mother partly explains why women, compared to men, often end up further down the scale at the end of their career. Figure 1.10 shows which part of the status attainment model is studied in Chapter 5. The main issue for this chapter is how the mother's occupational background, in addition to that of the father, affects the occupational career of the daughter. The last set of research questions read:

1. How do parents' role models affect the occupational career of the daughter?
2. How has the influence of the mother's occupation, compared to that of the father's, developed over time?

The empirical data used in this study had to fulfill two main requirements: they had to include a good measurement of the mother's educational and occupational status, the latter measured by a detailed code, in order to compile an occupational status score. For Chapters 4 and 5 the data had to cover respondents' first occupational title after leaving school and their full occupational careers. Such data are in fact quite rare. Where possible, I use cross-national data for replicational purposes by pooling them, to increase the statistical power and conceptual generalisation of the research design. Cross-national comparisons are not made. Due to the empirical restrictions encountered while work was in progress, two of the chapters (Chapter 3 and 4) nonetheless had to be confined to a national, Dutch perspective.

Most of the studies in this book cover an extended historical period to assess whether historical changes have taken place. Except for Chapter 3, all other chapters take this historical perspective on how the influence of the mother has developed over time. If the research question is geared towards a trend analysis, the extent and the direction of how the influence of the mother has changed historically can be studied by separately measuring her influence in subsequent cohorts.

The data used in Chapter 5 are so-called 'life history data'. The label 'life-history data' indicates, that for all respondents it is known when they finished their school, what their educational level was at the time they quit school, when they first entered the labour market, what their first occupational status was, when they quit their first employment and started their second job, what their next occupational status was, etc. It means that entire individual educational and occupational careers up to the time of the interview are mapped out. The advantage of life-history data compared with cross-sectional data is that we can study on an individual basis what causes some persons to have occupational transitions. A disadvantage of these data is that the further back in time career events have happened, the less the respondent is likely to remember these events correctly and place them into the right time frame. For Chapters 4 and 5, the data had to include a precise measurement of the respondents' first occupational status and in Chapter 5 their occupational career in addition. Hereafter follows a short overview of the data sets used in the empirical chapters.

This survey was carried out between 1992 and 1993 and was initiated by Ultee and Ganzeboom at the Department of Sociology at Nijmegen University (Ultee & Ganzeboom 1993). It contains a multi-stage random national sample of the Dutch population between 21 to 64 years. Included are 1000 primary respondents and 800 spouses, sampled from the community population registers. A probability sample was drawn from different Dutch communities, which were selected on the basis of their representativeness regarding their urbanization and region. In the Netherlands Family Survey 1992-1993 the entire socioeconomic characteristics and the family background of the respondents and their spouses are covered. The respondents gave full accounts on their life history, including their educational attainment as well as their occupational careers, entailing the timing of events, job titles and hours worked at the beginning and the end of a job spell.

This household survey was carried out in 1995 by the Utrecht Household Seminar at the Department of Sociology at Utrecht University, with as main investigators Weesie, Kalmijn, Bernasco, and Giesen (Weesie et al. 1995). The Households in the Netherlands 1995 contains 3354 respondents between 18 and 65 years, of which 1321 belonged to a panel study on the social integration of young adults (SI) and 2033 to the original Households in the Netherlands 1995 study. Couples are over-sampled, which means that the database includes more information on couples than single people, compared with the entire Dutch population. A probability sample was drawn from the address database of the national phone company (PTT afgiftebestand). This database from which the addresses were drawn also included unlisted people or people who had no phone in their home. The questionnaire of the Households in the Netherlands 1995 contained questions on the entire socioeconomic background of the respondent as well as life history data on their educational and occupational careers, in a similar fashion to that in the above study.

The first survey of the German Life History Study contained life history information for the birth cohorts 1929-31, 1939-41, and 1949-51 and was carried out in 1982 and 1983. The main initiator of this study was Mayer at the Max-Planck-Institute for Human Development in Berlin (Mayer 1983). The first survey contains 2171 respondents. The representativeness of the survey was secured in two steps. In a preliminary survey addresses and information on the number of households that would have to be contacted were obtained. In 420 electoral precincts a method called 'random-walk' was used to gather this information (every third household was contacted). In the second step the information obtained was compared with their representativeness in large household surveys. In the interviews subsequently carried out the respondents were asked about their socioeconomic background, and, again, gave full accounts on their life history covering their educational and occupational careers.The second survey of the German Life History Study contained life history information for the birth cohorts 1954-1956 and 1959-1961 (Mayer 1989). It contained 1008 interviews, with an average length of 67 minutes, which all were completed in 1989. The way the representativeness and addresses were secured for the second survey was slightly different from the first survey. Now the target population was selected from people who were listed in the public phone books, which had the disadvantage that anybody who did not own a telephone or was unlisted was not included in the study. It resulted in a slight under-representation of unemployed, apprentices, single people, and people from low income groups. The contents of the interview covered similar subjects to the first survey of the German Life History Study, only the target person and no other household members are included in the first and the second survey.

This household survey from the USA includes interviews with 13,017 respondents which were completed in the late spring of 1988. The main initiators were Bumpass, Sweet, McDonald, McLanahan, Sørensen and Thomsen at the Centre for Demography and Ecology at the University of Wisconsin, Madison (Sweet et al. 1988). The target population consists of the non-institutionalized adult population in the USA, 19 years and older, with an oversampling of minorities, one-parent families, families with stepchildren, cohabitors and recently married persons. The National Study of Families and Households is a national multi-stage area probability sample, drawn from 100 sampling areas in the USA. It also contains questions on the socioeconomic background and educational life history of the respondents, but it does not include questions on the occupational careers of respondents.

The education and the occupation of both parents serve as measure for the socioeconomic background of an individual. Educational levels were made comparable by approximating the number of years it would take the incumbent to reach a certain level. Appendix A of this study shows the procedure that was followed for the Netherlands, West Germany and the USA.

The present study uses the paradigm of socioeconomic status, not class, to investigate the contribution of the mother's occupation to status attainment. The concept of class has the disadvantage of having an aggregation level that is difficult to handle in an analysis of status transfer. Furthermore, the socioeconomic status tends to explain more of the variance in an empirical model than class does. Next, the scaling of occupations into continous socioeconomic classifications is simple to apply in empirical research. Finally, more consensus exists regarding the ranking of occupations into socioeconomic levels than of membership of people to classes (Sørensen 1994, Marshall et al. 1997, Grusky & Sørensen 1998). Throughout all the chapters the occupational codes (mainly ISCO or CBS occupational codes) of incumbents' jobs are scaled into the "International Socioeconomic Index of Occupational Status" (ISEI) for further use in the empirical analysis (Ganzeboom et al. 1992, Ganzeboom & Treiman 1996). Socioeconomic status scales are computed by using the education and the income of employed men to predict their occupational status. In some cases this procedure controls for the age of the incumbent. Other procedures exist to operationalize occupational status (e.g. Bose 1985, Wegener 1992). However, they seldom include an internationally comparative perspective for their status scales.

Occupational status scores which are based on the male employed population, such as the ISEI, have been criticized as not applicable to the female employed population, because females are usually paid less than males, even if they work in the same jobs. However, research comparing male- and female-based occupational scores shows that the correlations are very high, at 0.97 (Bose 1973). Considering the fact that male and female-based status scores seem to be very similar, it appears that the bias for the occupational status scores of the mothers in the data is negligible, given what is gained using internationally comparable measures such as the ISEI.

All chapters have been presented as separate articles at a conference, were published in a scientific journal, or were submitted for publication. Each chapter contains a full research report and can be read independently from the others. Although the organization of this study conforms to the logic of the status attainment model-dealing first with the education, thereafter with the first job, then with the career of the child-to a degree, the theoretical background of the chapters sometimes overlap. In Chapter 2 five different theoretical notions on how to measure the influence of parental socioeconomic background on children's education are empirically compared. Some additional ideas are proposed for the most efficient measure of parental background.

In Chapter 3 the focus is on the social consequences of mothers' employment for the educational level of her child, because mothers still carry the main burden of raising the children. The time restrictions and occupational resources of employed mothers are used to explain children's education.

In Chapter 4 we study the effects of the mother's occupational level on the first occupational status of her child, male and female, when he or she has finished school. Here we have extended the classical status attainment model to include the occupational sex-typing of the father, mother and child.

Next, Chapter 5 contains a study of the effects of parental occupational background on the daughter's occupational career. As results in Chapter 4 suggest that the mother's occupational background only affects her daughter's occupational attainment, sons are excluded in Chapter 5.

In the last chapter, Chapter 6, the conclusions from the previous empirical chapters are combined for every level of children's status attainment. General conclusions are drawn on the influence of the mother on the process of stratification and some challenges for future studies on social inequality expressed. Table 1.1 offers an overview on the prospective contents of the empirical chapters.

Footnotes:

1. 1962 was the year when the Occupational Changes in a Generation (OCG) survey was held.
2. Married women and men are, of course, a different population than fathers and mothers. Nevertheless, comparable population statistics for fathers and mothers are even more difficult to obtain than for married women and men. It would have been possible to use the population surveys of the empirical chapters of this book. Yet, sources other than my primary references underline the generalizability of the argument.
3. I thank Harry Ganzeboom for providing me with the data.
4. In the ILO statistics women were defined as belonging to the active female population if they were: (a) at work, performing some work for pay or profit during at least a specific brief period, either one week or one day; (b) with a job but not at work due to bad weather, strikes, illness, injury, vacation, etc.; (c) self-employed or unpaid family workers, working at least one third of the normal working time.
5. Furthermore there is no reason to assume that a priory only the level of the higher status parent counts whilst the lower status parent does not have any additional influence (see Chapter 2 for full details).
6. See also another study by Holland Baker (1984) who suggested that mother’s occupational status affects her daughter’s occupational status negatively; that is to the other extreme.