This article was downloaded by: [Radboud University Nijmegen] On: 23 June 2011 Access details: Access Details: [subscription number 907171856] Publisher Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37- 41 Mortimer Street, London WlT 3JH, UK t U R 0 P t. A N S 0 <.' I i -I ( i S European Societies Publication details, including instructions for authors and subscription information: http:/7www..informaworid.com/smpp/title~content=t713685208 EDUCATIONAL ATTA^IMENT, OCCUPATIONAL ACHIEVEMENTS, CAREER PEAKS Maarten H. J. Wolbers"; Ruud Luijkxb; Wout Ulteea Radboud University Nijmegen, Nijmegen, The Netherlandsb Tilburg University, Tilburg, The Netherlands Online publication date: 22 June 2011 1^- RTo cite this Article Wolbers, Maarten H. J., Luijkx, Ruud and Ultee, Wout(20U) 'EDUCATIONAL ATTAINMENT, OCCUPATIONAL ACHIEVEMENTS, CAREER PEAKS', European Societies, 13: 3, 425 - 450 To link to this Article: DOI: 10.1080/14616696.2011.568265 UKL: http://dx.doi.org/10.1080/14616696.20n.568265 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/tenns-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will-be complete or accurate or up to date. The accuracy of any instructions, formulae and dru9 doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. European Societies 13(3) 2011: 425-450 ©2011 Taylor & Francis ISSN 1461-6696 print 1469-8307 online Routledge Taylor&Francto Group EDUCATIONAL ATTA NMENT OCCUPATIONAL ACHIEVEMENTS CAREER PEAKS The Netherlands in the second part of the twentieth century Maarten H.J, Wolbers Radboud University Nijmegen, Nijmegen, The Netherlands Ruud Luijkx Tilburg University, Tilburg, The Netherlands Wout Ultee Radboud University Nijmegen, Nijmegen, The Netherlands ABSTRACT: This paper answers questions on the educational attainment and occupational career of men in The Netherlands whose working life began in the 1950s, 1960s, and 1970s, in so far as their job history is available until at least the age of 45 years. The analysis draws on five Dutch retrospective life-history surveys conducted between 1992 and 2003. The results show that a man's education depends upon his father's job, and that this effect has not changed for labour market entry cohorts. When explaining a man's first job, his father's job is influential once more, independent of a man's education. A man's education has a direct positive effect on his first job, his job after 10 and 20 years, and his peak job, but again the differences in status attainment between labour market entry cohorts are limited. Findings also reveal that advantages accumulate during a person's working life. Apart from a higher level of education, a higher first job has an independent positive effect on a man's job after 10 and 20 years, as well as on his peak status. Keywords: occupational career; social stratification; status attainment; The Netherlands Sociology has witnessed several generations of social mobility research (Ganzeboom et al. 1991). This paper avoids drawbacks of the fourth generation and brings in undervalued features of the second one. It answers questions on the educational attainment and occupational career of men whose working life began in the 1950s, 1960s, and 1970s, in so far DOI: 10.1080/14616696.2011.568265 425 EUROPEAN SOCIETIES as their job history is available until at least the age of 45 years. This paper's first section details generational shifts. The second portion presents a string of questions, the third one argues hypotheses, the fourth prepares data, and the fifth displays results. The final part reviews findings, recasts hypotheses and proposes follow-up questions. 1. Dropping poor questions and proposing a series of appropriate questions Successive generations of mobility research employed quite different techniques of data analysis. The first generation eyeballed tables classifying men's present job against their father's job (Lipset and Bendix 1959). The second generation estimated path models for the socioeconomic life cycle (Blau and Duncan 1967), and the third one fitted loglinear models to father-son class mobility tables (Goldthorpe 1980; Erikson and Goldthorpe 1992). Presently, event history models are en vogue, after their introduction by Blossfeld and Mayer (1988). In addition, every new technique of data analysis was accompanied by the collection of richer data. Each generation drew a representative sample from a country's population and took the job level of men as indicating their status in a society's stratification system. The Lipset generation made do with tables on a man's job and that of his father. The Duncan generation also ascertained a man's education and his job right after leaving school. The job of men 10 years on in their working life was added by the Goldthorpe (1980: 69) generation. Full occupational histories were collected by the Mayer generation, for men and women, as well as data for their father and mother. Given the increasing wealth of data and the growing sophistication of their analysis, it may be surmised that early generations were interested in descriptive issues, later ones in comparative questions, and recent ones in explanatory problems. Yet each generation described and explained. Later generations deleted questions that 'pose the issue poorly' (Blau and Duncan 1967: 402). This often-overlooked aspect of generational shifts deserves top billing and is elucidated now. By subtracting father's job from son's job, the first generation ascertained how many men were upwardly mobile, downwardly mobile and stationary. However, the second generation discarded the question of why some men move up and other men down. It demonstrated mathematically that a weak association between a man's education and the extent to which he ascended or descended compared with his father need not be surprising. Given a strong association between a man s education and his job, the stronger the association between father's job and son's education is, the weaker the association between education and the 426 Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE mobility variable will be. Sons with more education have a higher job father to begin with - and because of this ceiling, only a few will ascend. Something similar holds for any association involving a difference score (Blau and Duncan 1967: 194-9). The second generation barred difference scores, and went for associations between elementary variables. Questions should involve the strength of the association between the status of persons and that of their parents. In this way, the con question of the effect of education on mobility is replaced by two right-on-target questions: that of the effect of father s occupation on son's education, and that of the consequences of father's job and son's education for son's job. Like the second generation, the third one settled for associations between elementary variables. However, it held that jobs cannot be arranged along one continuum, as path models assume, but belong to a limited number of categories. It also held that, even if a gradient was apposite, the strength of the association between parental and personal status may differ for sections of the breath of jobs. An example is more mobility in the middle than at the top and bottom. So, path models were not attuned to pertinent questions, and hence the third generation s interest was in the unequal outcomes of competitions between persons from different origins for higher and lower destinations, as indicated by odds ratios (Goldthorpe 1980; 77). Odds ratios are measures for the association between categorical variables, and the parameters of the log-linear models estimated in third generation research can be reworked into odds ratios. The fourth generation held that the questions of earlier ones seem underspecified, and actually are compound questions that cannot be broken down into doable ones. Mobility questions should specify two points in time. The data of earlier generations do refer to the same year for every man's current job, but pertain to different dates for their father's job. If this is the case, origin effects will depend upon the length of the period considered. After all, experience might speed up a man's job level. This spoils the possibility of answering questions about the separate influence of factors like educational reforms and unemployment levels on current job. More generally: age, cohort, and period effects cannot be identified. That is why the fourth generation did away with questions on the association between origin and current job and ascertains complete careers. By explaining a person's job at the close of always equally short spells, questions sorting out the various kinds of effects can be answered. Depending upon the histories collected, this period is measured as a month or as a year. 427 EUROPEAN SOCIETIES 2. Models for the socio-economic life cycle reappraised and a string of questions proposed One substantive argument against second generation path models has been their unimportance for sociology. In a graph with dots and arrows (Blau and Duncan 1967: 170), dots stand for properties of persons, and no property refers to the societies persons belong to (Lenski 1984: xvi). However, arrows do refer to societal features. For instance, one arrow stands for the status returns to an additional year of education, in a specific country at a certain moment. This yield was found to vary within one society across cohorts (Blau and Duncan 1967: 178), and comparisons of the strength of associations between elementary variables for cohortscountries are blossoming (Rijken 1999). So, we beg to differ. Indeed, an underappreciated advantage of the second generation is that it addresses questions about societies in a neat sequence. It replaced the first generation question of the across time and between societies varying effect of origin on destination (in the singular) by three questions: the effect of father's job on son's education, the effect of father's job (net of that of son's education) on son's first job, and the effect of father s job (net of that of son's education and first job) on son's current job. Against this background, this paper deals with a series of questions, with the first question referring to the earliest phase of the status attainment process, and later ones to successive stages and specific features of the whole process. This paper's first question (Ql) is: Did a. higher family background for men starting to work in the 1950s, 1960s, and 1970s make for more education, and was this effect the same for all three cohorts? Our second question does not focus the outcome of a man's educational career, but the start of his occupational career (Q2): Did, for each cohort, the effect of a man's education on his first job increase, while that of his origin, net of education, decreased? If a man's job history is available, it is easy to replace the off-the-mark second generation questions about a man's present job by look-alike right ones. Our third question {Q3) is on effects of background and education on a man's job 10 and 20 years after labour market entry: Are effects of background and education on a man's job after 10 and 20 years so powerful, that they remain present after taking into account their effects on a. man s first job, or do background and education achieve their results in one go? Path models have shown that a man's current job is higher if, independent of his education, his first job was higher. This neglected result is substantively interesting. Careers seem self-reinforcing: among men with the same education, those starting in higher jobs finish higher than those who begin in lower ones. This paper's fourth question (Q4) is: 428 Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE To what extent is a man's job 10 and 20 years on in his working life higher, if his first job was higher, net of the effect of education and background? Older men have had more chances to find a job matching wishes and skills than younger men. But then, after finding it, they risked losing it. The Netherlands is a market economy with varying unemployment rates. So, a fifth question (Q5) is on the spot: Does a higher unemployment rate lower the level of a man's first job, and his job 10 years and his job 20 years on in his working life? Our two final questions deal with fourth generation drawbacks. By breaking job histories down into periods of equal length, this generation bypassed common sense features of careers: their progress and their peak. This paper deals with progress by highlighting first jobs, jobs 10 and jobs 20 years later. To bring in peaks, our sixth question (Q6) is about a man's job during each and every month of his working life below his peak. Do background, education, time on the labour market, and unemployment rate affect a man s chances of remaining below peak or arriving on top? Our seventh question (Q7) recognises that peaks may differ. Its starting point is the highest job reached by persons 45 years or older. Does the level at which a- man's career peaks depend not only on his background and education, but also on his first job and the time he was on the labour market? This question need not be about a man's job at age 45, since a man may peak before this age and slide after peaking. 3. Hypotheses: families, markets, and states The Duncan generation has been heavily criticised from substantive points of view. The terms may have changed, but the old argument that 'status attainment research' is 'a-theoretical' stands behind its current name of abuse 'variable sociology' (Hedstrom 2005). However, Horan (1978) pinpointed quite specific hypotheses. He asserts effects of changes in a society's prime mode of subsistence: in societies like The Netherlands, the shift from agriculture, by way of manufacturing, to knowledgeintensive services was accompanied by a change in the norms regulating status attainment. Whereas agrarian norms stipulated ascription (following in father's footsteps), later ones prescribe achievement. Higher jobs go to persons with educational credentials, and education becomes independent of origin. Of course, this from-ascription-to-achievement hypothesis may falter in research. However, it also may be the case that it faces obvious difficulties, or that more general theories imply its falsify. We discuss these points now. 429 EUROPEAN SOCIETIES To begin with, without surveillance norms wither away (Tittle 1980), The judiciary enforces laws about property, but there is no equivalent when it comes to fair decisions after job applications. Surely, profits suffer if firms reject applicants with better credentials, but companies sometimes are in the red. So, the chances that higher jobs go to persons with more education may remain unchanged. Secondly, behind the from-ascription-to-achievement hypothesis stands the idea of education as a functional requirement of industrial societies (Kerr et al, 1960). This is an obscure notion. Persons have unsatisfied needs, so why would the needs of societies, if they have them, become fulfilled? A pervasive trend towards smaller effects of background on education may be absent too. Thirdly, when 'positive externalities' of education are highlighted, a new difficulty arises. The theory of collective goods (Van den Doel 1979) states that markets do not optimally produce positive externalities, and The Netherlands is a market economy. Yet it also holds that states sometimes provide collective goods from taxes. Now, schooling is compulsory in The Netherlands, the school-leaving age was raised, and state-funded stipends for post-compulsory schooling are considerable. The hypothesis answering this paper's question about educational attainment (Ql) therefore reads that, at least for The Netherlands in the period considered, for later cohorts origin effects on education are smaller than for earlier ones. Finally, this hypothesis comes with a rider. Educational reforms may have done away with effects of 'material resources', but these reorganisations brought in consequences of 'cultural resources'. Bourdieu's (1979: 177) theory of compensatory strategies posits that higher class parents compensate for state-funded stipends, by transferring more cultural resources to their children. So, origin effects will not vanish. Since the mid-1980s, when youth unemployment soared, Dutch governments invest heavily in easing the transition from school to work (Visser and Hemerijck 1998). The answer to this paper's question about the possibly changing impact of a man's education on his first job (Q2) states that the effect of a man's education on his first job increased, whereas the net effect of origin decreased. Again, net effects of background will not have disappeared. Higher origins make for the possession of less tractable and more tacit resources valuable within corporations too. When applying for jobs, persons state their education and experience. Although employers will take experience of persons into account, they will not bypass credentials. So, to answer our question on non-mitial jobs (Q3), we postulate net effects of a man's education on his job after 10 and 20 years. The same reasoning answers our question on first job effects (Q4): a 430 Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE a >, man's job 10 years after his first one will be higher if his first job was higher, and so will his job after 20 years. This paper's next question is about effects of aggregate unemployment rates. The Netherlands witnessed full employment until the 1970s. With the two global oil crises, the business cycle resurfaced. In the 1980s the decline of manufacturing took a big toll. In those years, the growth of the Dutch economy was at its post-war low. Hence, the hypothesis pertinent to our unemployment rate question (Q5) postulates independent effects of the unemployment rate on job level. We now answer our last but one question on factors affecting career peaks. The prime hypothesis answering it invokes the time a person has spent on the labour market. On the one hand, hypotheses about technological exigencies hold that persons almost start at their peak. On the other hand, hypotheses about fortuitous markets predict that the time it takes for persons to reach their peak may be quite long. Since the size of each effect is unclear, it cannot be stated how they add up. But then, if there is an effect of unemployment rates, given their trend in The Netherlands, it should be stronger for later cohorts than for earlier ones. Our final question is not about factors making for a peak, but about the level of this peak. Here our hypotheses are straightforward. Higher current unemployment rates make for a lower peak in a man's occupational career, more education leads to a higher peak, as does a higher origin and a longer working life. 4. Research design 4.1. Data In this paper we analyse data from five retrospective surveys: Netherlands Family Survey 1992-1993 (Ultee and Ganzeboom 1992), Households in the Netherlands 1995 (Weesie et al. 1995), and Family Surveys Dutch Population 1998, 2000, and 2003 (De Graafe? al. 1998, 2000, 2003). Each survey involves a random sample from all Dutch inhabitants and face-toface interviews with respondents at home. The number of respondents in the surveys was 1,800, 3,354, 2,029, 1,561, and 2,174, respectively, yielding in total 10,918 respondents. The analyses to be presented refer to men, since the few women with a complete job history would be atypical. We deleted men below the age of 45 years at the moment of survey. We deemed it unwise and unnecessary to select only men who have definitively stopped working, and opted for the age of at least 45 years to 431 EUROPEAN SOCIETIES be pretty sure that men have reached their career peak, Indeed, one third generation study compared cohorts by assuming that the occupation of men had matured around the age of 35 years (Goldthorpe 1980). Nevertheless, the choice for men at least 45 years old severely reduced the number of cases for modelling. After list-wise deletion of respondents for whom information is missing on any of the variables, a set of 1,087 men remained. We lost about half of our original samples by selecting men only, and more than half of these men by deleting those below the age of 45 years. We return to our decision to restrict our analysis to men in this paper's final section. The surveys ascertained all the jobs persons held in the past, although with some difference in detail. For each job held by a respondent, the start and end dates are reported, as well as features of the job. In the Netherlands Family Survey 1992-1993, work histories were organised by job spells, in the other surveys by employer spells. Within each employer spell, information was gathered on the jobs held. For the survey Households in the Netherlands 1995, this was limited to the first and last job. 4.2. Measurements Job level is measured by assigning to each job title a score from the International Socio-Economic Index of Occupational Status (ISEI) (Ganzeboom et al. 1992). This scale ranges from 16 for occupations with the lowest to 90 for those with the highest status. In the analysis, five jobs are considered: the job of the father when the respondent was 15 years old, the first job of men, their job after 10 years, the job after 20 years, and their peak job. The highest level of education attained by men comprises six categories for the most distinct qualifications awarded by Dutch schools: elementary education (lo), lower secondary education (Ibo/mavo), higher secondary vocational education (mbo), higher secondary general education (havo/vwo), lower tertiary education (hbo), and higher tertiary education (wo). As stated, three labour market entry cohorts are distinguished: the 1950s, 1960s, and 1970s. Yearly unemployment rates are taken from Statistics Netherlands (CBS 2007). Work experience is measured as the number of months in employment since entering the labour market. We model a possible curvilinear effect of work experience by including both a linear and quadratic term. 432 WOLBERS, LUIJKX & ULTEEStatus attainment in The Netherlands 5. Results 5.1. Educational attainment Table 1 contains multinomial and ordered logit models for a man s highest level of education. Model 1 reveals that men from recent cohorts have more education than men from earlier ones. In addition, it shows that father's status increases his son's education. Model 2 divulges that the effect of father's job remained constant over time. This result speaks against the hypothesis of smaller effects for later cohorts, which answered Ql. 5.2. Occupational achievement at labour market entry Table 2 presents linear regression models for the status of a man's first job. Model 1 shows that men from recent labour market entry cohorts started in jobs with a higher status score. For instance, men who started their career in the 1970s have a job almost seven points higher than those who started out in the 1950s. Furthermore, the model tells that father's job status positively affects his son's job status. The unemployment rate at the time of labour market entry does not affect the level of a man's entry job. Model 2 shows the importance of education in the allocation of men to jobs. The R increases from 0.139 to 0.380. High educated men entered the labour market in much higher jobs than low educated ones. The difference between the most and least qualified men (the contrast between those with primary school only and university graduates) amounts to 23 status points. In addition, Model 2 reveals that cohort effects disappear when taking account of educational differences. So, the advantageous position of recent cohorts in terms of the status of their first job can be attributed to their, on average, higher qualifications at labour market entry. The direct effect of father's status on his son's status is reduced considerably, once controlled for educational qualifications. Around half (that is, 49 (1-0.161/0.316) percent) of the status inheritance gets channelled by education. None of the interactions between education and father's status on the one hand and cohort on the other is significant (Model 3). So, the direct effect of this achieved, respectively, ascriptive property at labour market entry did not change over time, which undermines our hypothesis regarding this paper's question on first jobs (Q2). Nevertheless, at this stage achievement overshadows ascription, given the stronger impact of education on a man's first job, compared with that of father s status. 433 s TABLE 1. Determinants of highest level of education attained (multinomial and ordered logit analysis) ^.33 ? 8 w Level of education Multinomial Lbo/mavo Mbo Havo/vwo Hbo Wo Ordered Model 1 Labour market entry cohort (1950s = ref.) 1960s 1970s Status father's occupation (ISEI) Constant Constant (cut 1) Constant (cut 2) Constant (cut 3) Constant(cut 4) Constant (cut 5) Model x2 df Number of men Model 2 Labour market entry cohort (1950s = ref.) 1960s 1970s Status father's occupation (ISEI) * 1960s * 1970s Constant Constant (cut 1) Constant (cut 2) 0.275 1.096 0.009 1.075** 208** 15 1,087 1.086 1.463 0.005 0.019 0.069 0.893 0.481 2.090** 0.003 0.630 2.064** 1.667 0.019 0.038- 0.016 0.274 0.429 2.475** 0.043** 3.050** 1.260 0.098 0.047* 0.019 0.061 -3.180** 0.531* 2.446** 0.023** - 0.502 1.203 0.631 0.029* -0.016 0.050 - 0.714 0.949** 3.219** 0.054** 3.169** 0.926 0.044 0.043** 0.003 0.078 2.588** 0.410** 1.260** 0.033** 0.584** 1.031** 2.118** 2.360** 3.984** 172** 3 1,087 0.269 0.861* 0.029** 0.003 0.009 0.741** 0.869** -p> Cd U1 »m ffl I TABLE 1 [Continued] Level of education Multinomial Lbo/mavo Mbo Havo/vwo Hbo Wo Ordered 60 120 180 240 300 360 Primary and lower secondary -- Higher secondary Tertiary Rgure 2. Percentage having reached career peak for different labour market entry cohorts (left part) and levels of education (right part) by working experience (in months). TABLE 4. Determinants of reaching career peak (discrete-time event history analysis) and occupational status of job at career peak (linear regression analysis) w m Reaching career peak Status at career peak Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 ^co Labour market entry cohort (1950s = ref.) 1960s 1970s Status father's occupation (ISEI) *1960s *1970s Level of education (Lo = ref.) Lbo/mavo * 1960s *1970s M bo * 1960s * 1970s Havo/vwo *1960s *1970s Hbo "1960s * 1970s Wo * 1960s *1970s Status first occupation (ISEI) * 1960s * 1970s 1.295* 5.775** 0.055** 1.144 0.576 2.502* 4.297** 9.203** 2.129** 5.667** 0.036** 1.396 1.220 3.859* 4.703** 9.915** 0.282** - 2.280 -9.518* 0.016 -0.012 - 0.024 -1.139 -0.176 3.682 -1.316 2.101 10.606* -1.491 3.924 4.439 -1.528 2.487 7.978 - 1.350 7.274** 5.047 0.222** 0.043 0.220 2.255** 1.770 0.111** 1.667 6.204** 11.953** 15.506** 21.117** 1.349 1.167 0.038 2.083 5.248** 8.415** 9.215** 11.466** 0.501** 7.842* 0.653 0.054 0.022 - 0.047 2.000 - 0.008 4.907 5.161** -0.829 6.827 6.659* -0.704 12.929* 9.511** -0.701 5.813 7.329** 4.236 13.308* 0.579** -0.122* 0.105 01 I ^ Iw g .S3 w 8° c: m m t .J3 ? 8 TABLE 4 {Continued) m 0) Reaching career peak Status at career peak Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Working experience Working experience squared Current aggregate unemployment rate Constant lnsig2u constant Model x2 df R2 Number of men Number of man-months 0.078"-* -0.000** - 0.081* -16.537** 4.567** 1,245** 11 1,086 134,684 0.100** -0.000** - 0.033 -28.039** 4.804** 918** 12 1,086 134,684 0.074** -0.000** - 0.069 -21.107** 4.159** 1,691** 26 1,086 134,684 0.027** 0.000 -0.067 41.910** 0.325 1,086 0.078** 0.078** -0.000** -0.000** - 0.003 - 0.026 22.239** 18.607** 0.520 1,086 0.529 1,086 *P< 0.05, **P<0.01. Note; 'lnsig2u constant" denotes the log of the panel-level variance of the intercept. Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE men of the 1950s cohort. This supports the hypothesis that technological developments make for quick peaking (part of our answer to Q6). Men with a high status father are more likely to peak, too. The education effects indicate that during a certain month men with higher secondary general education and tertiary education have higher chances of attaining their maximum than less educated ones. Furthermore, men with more work experience are more likely to reach their top than less experienced men. In addition, in times of high unemployment chances of peaking are below those in times of low unemployment. Results remain unchanged when a man's first job is taken into account (Model 2), with the exception of the origin effect and the unemployment rate effect. Upon introducing the level of a person's first job, the effect of the current unemployment rate upon reaching a peak vanishes. The origin effect now is negative. Perhaps some ceiling effect, so feared by the Duncan generation, shows up here. A person's first job has a strongly positive effect on reaching the top. Model 3 shows two significant interactions. They pertain to education and cohort. First, men with higher secondary vocational education reached the top relatively more often when they started working in the 1970s. Second, men with higher tertiary education relatively more likely reached their career peak in a certain month when they belonged to the 1960s cohort. Linear regression models for a person's peak status are presented in Table 4 as well. Model 1 shows significant differences in peak status between the 1950s and the 1960s labour market entry cohorts, after taking origin, education and first job status into account, but no significant differences between the 1950s and the 1970s cohort. So, of the three cohorts, the 1960s one fared best. Its peak status was two points higher. Furthermore, Model 1 tells that father's status matters. The peak in the career of advantaged origin sons is higher than that in the career of men from low origins. Education has a strong positive impact on maximum status. However, at this stage of a man's career the education effect seems a bit weaker than at labour market entry. For the first job, the difference in status between men with the highest and the lowest education was 23 points (Table 1). For the peak job it is 21 points. Model 2 adds a man's first job, and it strongly affects maximum status, Moreover, it to a large extent mediates the effect of education. In Model 1, the difference in status points between the least and the most qualified men was 21 points; in Model 2, it is 11 points. This implies that half of the direct effect of education can be statistically explained by the status of a man's first job. The effect of father's occupation is interpreted to a strong degree by the status of the first job as well: the significant impact of 0.111 of father's occupation in Model 1 is reduced to an insignificant 0.038 in 445 EUROPEAN SOCIETIES JVlodel 2. Furthermore, the effect of working experience is significant after controlling for first job. Men with more working experience have a higher job at their career peak than those with less experience. Finally, the effect of the 1960s cohort compared with the 1950s cohort becomes insignificant. The 1960s cohort apparently had higher labour market entry jobs. All in all, these findings confirm our hypotheses regarding the level of a person's peak (and answer our Q7). Finally, Model 3 includes interactions of cohorts with origin, education and first job. Three interactions are significant, and the significant main effect in Model 1 for the 1960s cohort compared with the 1950s cohort shows up again. Once again, people from the 1960s cohort have a higher peak, this time by 8 points. Moving on to the significant interactions, intriguingly, for men from the 1960s cohort, the level of their first job had a smaller effect on their peak status than for men of the 1950s cohort. The other two significant interactions pertain to education variables. The effect of the highest level of education (wo), compared with the lowest (lo), on status at career peak is stronger for the 1970s cohort, and the same goes for the effect of the highest but two level of education (havo/vwo). It is puzzling why the highest but one level (hbo) does not show a significant interaction. 6. Discussion This paper broke down the question of the effect of origin on destination, which occupied various generations of mobility research, into a series of questions. It raised questions about effects on a men's level of education, the level of their first job, the level of their job 10 and 20 years later, and their peak job. Each time, this paper raised questions about effects of parental background, a man's education, and his previous jobs. The questions pertained to men entering the Dutch labour market in the 1950s, 1960s, and 1970s, in so far as their job history has been ascertained until the age of 45 years. A man's education, of course, was found to depend upon his father s job, and it also affected his own first job. When explaining a man's first job, his father's job was influential once more, independent of a man s education. Apart from that, a man's education directly affected his first job, his job after 10 and 20 years, and his peak job. However, the job of a man's father neither directly affected his job after 10 and 20 years, nor his peak job. These results show the importance of'recurrent effects', but also underline the limits of this notion. Results also show that advantages accumulate during a person's working life. A higher level of education made for a higher first job and a higher job 446 Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE after 10 and 20 years, but a higher first job also had an independent positive effect for a man's job after 10 and after 20 years, as well as for his peak status. Therefore, hypotheses on 'cumulative advantages' deserve elaboration (DiPrete and Eirich 2006). Our finding that the effect of a man's first job for his job after 10 years is stronger than the effect of his first job for his job after 20 years, provides a pointer. This paper found limited differences between cohorts. The effect of origin on education was stable. Later cohorts did have higher job levels at the outset of their career, but not after taking father's job level into account. We found a direct effect of father's job on a man's first job, but it did not differ from cohort to cohort. The effect of education on a man's first job did not differ across cohorts either. When explaining a man's job after 10 years, after 20 years, and a man's peak job, after taking into account the effect of father's job on first job and of first job on later job, no direct effects of father's job was found. One theoretical implication of our findings therefore is that the various generations of research have spent too much time on changes supposedly induced by industrialisation and states. We did find that the effect of a man's first job on his job 10 years later was smaller for the 1970s cohort than for the 1950s cohort. We found other cohort differences in effects too. Effects of two out of five contrasts for level of education on the chances of career peaking turned out to differ across cohorts, and so did two out of five contrasts for education on peak status. However, these effects did not pertain to the same contrasts and the same cohorts. The second generation of mobility research went after one measure for associations between ascribed and achieved properties of men, while the third generation aimed for the more detailed characterisation of an association. Here we applied this point not to the effect of father s job on son's job at various moments of his working life, but instead we used it, with some success, to determine the consequences of a man's educational qualifications for various stages in their occupational career. The result of limited differences of effects between cohorts may seem at variance with other findings for The Netherlands. However, these studies stacked data-sets more heterogeneous than the ones analysed here. They also pertain to more and other cohorts than the ones considered here (De Graaf and Ganzeboom 1990) and lump men early in their career together with men at the end of their career (De Graafand Luijkx 1992). Indeed, findings on trends always should invoke the period studies. Although the present paper estimated quite a few models, it may be held that we skipped some obvious ones. For instance, we presented the effect of the aggregate unemployment rate on a man's entry job, his job after 10 and 20 years and his peak job, but bypassed the issue of whether individual unemployment at labour market entry forms an obstacle for the 447 EUROPEAN SOCIETIES later career (Luijkx and Wolbers 2009). We did not do so, since we found no effect of the unemployment rate on entry job, job after 10 years and peak job. Another model we did not present ascertains the effect of a man's job 10 years after entering the labour market on a man's job another 10 years later. We here state the result of this model: a man's job after 20 years is not only higher if this man's level of education is higher, but also if this man's entry job was higher, and, net of this, if this man's job after 10 years was higher. This is additional evidence for the notion that a man's occupational career is a self-reinforcing process. We skipped this model so as not to spoil a test of hypotheses about changing effects of a man's education and of his first job. One question the present paper did not address was that of career dips for men. It deserves separate attention. In addition, until now no neat series of questions on women, stratification and mobility has been addressed for The Netherlands. Indeed, already the wording of a neat series of questions on this theme is an exercise on its own. Questions should not only seek to explain job levels, but also (non)employment, and they should incorporate not only effects of family background, but also of spouses (Bernasco et al. 1998). References Bernasco, W., De Graaf, P. and Ultee, W. (1998) 'Coupled careers', European Sociological Review 14: 15-31. Blossfeld, H.-P. and Mayer, K.-U. (1988) 'Labour market segmentation in the Federal Republic of Germany', European Sociological Review 4: 123-40. Bourdieu, P. (1979) La distinction, Paris: Minuit. Blau, P. and Duncan, 0. (1967) The American Occupational Structure, New York: Wiley. CBS (2007) Historic geregistreerde werkloosheidscijfers, Voorburg/Heerlen: Statistics Netherlands, http://statline.cbs.nl De Graaf, P. and Ganzeboom, H. (1990) 'Intergenerational educational mobility in the Netherlands for birth cohorts from 1891 through I960', The Netherlands, Journal of Social Sciences 26: 35-50. De Graaf, P. and Luijkx, R. (1992) 'Van "ascription" naar "achievement"? Trends in statusverwerving in Nederland tussen 1930 en 1980', M.ens en Maatschappij 70: 412-33. De Graaf, N.-D., De Graaf, P., Kraaykamp, G. and Ultee, W. (1998) Family Survey Dutch Population 1998 [machine-readable data file 448 Status attainment in The Netherlands WOLBERS, LUIJKX & ULTEE Pl 5 83], Department of Sociology, Radboud University Nijmegen [producer], The Hague: Dans [distributor]. De Graaf, N.-D., De Graaf, P., Kraaykamp, G. and Ultee, W. (2000) Family Survey Dutch Population 2000 [machine-readable data file P1609], Department of Sociology, Radboud University Nijmegen [producer], The Hague: Dans [distributor]. De Graaf, N.-D., De Graaf, P., Kraaykamp, G. and Ultee, W. (2003) Family Survey Dutch Population 2003 [machine-readable data file P1792], Department of Sociology, Radboud University Nijmegen [producer], The Hague: Dans [distributor]. DiPrete, T. and Eirich, G. (2006) 'Cumulative advantage as a mechanism for inequality', Annual Review of Sociology 32: 271-97. Erikson, R. and Goldthorpe, J. (1992) The Constant Flux, Oxford: Clarendon Press. Ganzeboom, H., Treiman, D. and Ultee, W. (1991) 'Comparative intergenerational stratification research', Annual Review of Sociology 17: 277-302. Ganzeboom, H., De Graaf, P. and Treiman, D. (1992) 'A standard international socio-economic index of occupational status', Social Science Research 21: 1-56. Goldthorpe, J. (1980) Social Mobility and Class Structure in Modern Britain, Oxford: Clarendon Press. Hedstrom, P. (2005) Dissecting the Social, Cambridge: Cambridge University Press. Horan, P. (1978) 'Is status attainment research atheoretical?', American Sociological Review 43: 534-41. Kerr, C., Dunlop, J., Harbison, F. and Myers, C. (1960) Industrialism and Industrial Man, Cambridge, MA: Harvard University Press. Lenski, G. (1984) Power and Privilege (2nd edition), Chapel Hill, NC: University of North Carolina Press. Lipset, S. and Bendix, R. (1959) Social Mobility in Industrial Society, Berkeley, CA: University of California Press. Luijkx, R. and Wolbers, M. (2009) 'The effects of non-employment in early work life on subsequent employment chances of individuals in the Netherlands', European Sociological Review 25: 647-60. Rijken, S. (1999) Educational Expansion and Status Attainment, Amsterdam: Thela. Tittle, C. (1980) Sanctions and Social Deviance, New York: Praeger. Ultee, W. and Ganzeboom, H. (1992) Netherlands Family Survey 1992-1993 [machine-readable data file P1245], Department of Sociology, Radboud University Nijmegen [producer], The Hague: Dans [distributor]. 449 EUROPEAN SOCIETIES Van Den Doel, H. (1979) Democracy and Welfare Economics, Cambridge: Cambridge University Press. Visser, J. and Hemerijck, A. (1998) A Dutch Miracle. Job Growth, Welfare Reform and Corporatism in the Netherlands, Amsterdam: Amsterdam University Press. Weesie, J., Kalmijn, M. and Ganzeboom, H. (1995) Households in the Netherlands 1995 [machine-readable data file P1458], Department of Sociology, Utrecht University [producer], The Hague: Dans [distributor]. Maarten H.J. Wolbers is an Associate Professor in the Department of Sociology, Radboud University Nijmegen. He previously worked at the Research Centre for Education and the Labour Market, Maastricht University and the Department of Social Research Methodology, VU University, Amsterdam. His research interests include the sociology of education, work and labour markets. Ruud Luijkx is an Associate Professor in the Department of Sociology, Tilburg University. His research focuses on the comparative research of educational outcomes, social mobility and life-courses. Wout Ultee defended his Ph.D at the University of Utrecht in 1977, and has held a chair in general sociology at the Radboud University Nijmegen since 1987. He has published with various Dutch sociologists on educational heterogamy in 23 highly industrialized countries and 67 countries from all over the world, coupled careers in the contemporary Netherlands, social mobility and voting in the contemporary Netherlands, trends in religious heterogamy in Germany and The Netherlands, and trends in the relation between religion and suicide in The Netherlands. Address for correspondence: Maarten H.J. Wolbers, Department of Sociology, Faculty of Social Sciences, Radboud University Nijmegen, P.O. Box 9104, 6500 HE Nijmegen, The Netherlands. E-mail: rn.wolbers@maw.ru.nl 450