Toward Open Societies? Trends in Male Intergenerational Class Mobility in European Countries during Industrialization1 Ineke Maas and Marco H. D. van Leeuwen Utrecht University Do the observed increases in intergenerational mobility in European societies in recent decades have their origin in 19th-century industrialization, as is posited by the industrialization thesis? Using over 600,000 marriage records and an internationally and historically comparative measure of occupational class, the authors study total and relative intergenerational mobility of men in Britain, Finland, France, Germany, Hungary, the Netherlands, and Sweden between 1800 and 1914. For these countries together and for most countries separately the preindustrial period was characterized by stable or decreasing total and relative mobility, whereas a trend toward greater mobility took place during industrialization, lending qualified support to the industrialization thesis. INTRODUCTION The degree to which men and women can escape their class of birth during their life course is a perennial question in both the social sciences and public 1 This research was supported by the European Research Council (Advanced Grant no. 230279, 2009–14), “Toward Open Societies? Trends, Variations and Driving Forces of Intergenerational Social Mobility in Europe over the Past Three Centuries.” We thank the participants at the RC28 meeting in Trento, 2013, the MaSS seminar at Utrecht University and the SILC seminar at the VU Amsterdam, and the AJS reviewers for their helpful comments. And we are especially grateful to all researchers who made their data available to us. Direct correspondence to Ineke Maas, Department of Sociology/ ICS, Utrecht University, Padualaan 14, 3584 CH Utrecht, the Netherlands. E-mail: [email protected] © 2016 by The University of Chicago. All rights reserved. 0002-9602/2016/12203-0005$10.00 838 AJS Volume 122 Number 3 ( November 2016): 838–885 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility discourse. In closed societies those born into lower classes have few opportunities to improve their position, regardless of how hard they try or how talented they are. Closed societies are often thought to be a phenomenon of the past, of which only a few remnants have survived. The Industrial Revolution, starting in Britain and spreading over Europe during the 19th century, is often viewed as having been the starting point of a trend toward both more total mobility (i.e., an increase in the percentage of men or women in a class different from that of their parents) as well as more relative mobility (i.e., more equal relative chances of men and women from different classes of origin to reach certain classes of destination) (Inkeles 1960; Kerr et al. 1960; Blau and Duncan 1967; Form 1979). In this study we investigate this claim. We ask how total and relative mobility from father to son in European societies changed as these countries developed from preindustrial to industrial societies. Our focus is thus on the first countries that industrialized (countries that industrialized later, in an already partly industrialized world, are not the topic of the original theoretical claim and may show different effects of industrialization). Our focus is on men only. Although we use very rich data, there is not enough information on women’s class position to include women in the analysis. Our study is a new addition to the research tradition in which changes in mobility patterns are studied in several countries against the background of historical change, however, without explicitly modeling the effects of these changes. Several country-specific studies have shown increasing total or relative mobility for the second half of the 20th century (e.g., Erikson [1983] and Jonsson and Mills [1993] for Sweden, Goldthorpe and Portocarero [1981] and Thélot [1982] for France, Wong and Hauser [1992] for Hungary, and Ganzeboom et al. [1987] for the Netherlands). But in other countries such a trend has not been found (e.g., Erikson, Goldthorpe, and Portocarero [1983] for Britain, and Ishida, Goldthorpe, and Erikson [1991] for Japan). Erikson and Goldthorpe [1992] investigated changes in total and relative mobility comparatively in a range of countries. In 1992 they published The Constant Flux: A Study of Class Mobility in Industrial Societies, the result of more than 10 years of studying trends and country differences in total and relative mobility (see, e.g., Erikson et al. 1979, 1982, 1983). The aim of The Constant Flux was to describe rates and patterns of intergenerational class mobility in industrial societies, with special reference to the extent of change and variation, constancy, and commonality (Erikson and Goldthorpe 1992, p. 2). Using data collected in the 1970s, they compared nine countries (England and Wales, France, the Federal Republic of Germany, Hungary, the Irish Republic, Northern Ireland, Poland, Scotland, and Sweden). Birth cohorts were compared in order to allow conclusions to be drawn about changes in mobility between approximately 1935 and 1970. Erikson and Goldthorpe studied both total and relative 839 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology mobility. They concluded that there was no evidence of general trends toward higher levels of total mobility or relative mobility, nor evidence that relative mobility rates were changing in any consistent direction, nor evidence that these rates showed a tendency over time to become more similar cross-nationally. In other words, mobility patterns were in a constant flux. The conclusions of Erikson and Goldthorpe were contested even before their study had been published. In a study of 35 countries, Ganzeboom, Luijkx, and Treiman (1989) found major differences in relative mobility between countries and a trend toward greater openness between 1947 and 1987. They concluded that the relative mobility of men had increased slowly and that this increase was a global phenomenon. The fact that the two groups of researchers reached different conclusions is often explained by differences in data and method. Ganzeboom et al. not only included more countries in their study, they also analyzed several mobility tables per country. The chronological spread of these tables allowed them to draw conclusions about trends without having to rely on birth cohorts. Another possible explanation is that their study covered a period that ended 15 years later than that studied by Erikson and Goldthorpe and thus for example included the economic crisis of the early 1980s.2 Based on a reanalysis of a subset of the data of Ganzeboom et al. (1989), Wong (1992) doubted whether the increase of relative mobility was as global as suggested. Social Mobility in Europe (Breen 2004) appeared as a followup to the study by Erikson and Goldthorpe for a more recent period, covering the early 1970s to the late 1990s. Eleven countries are compared: Britain, France, Ireland, West Germany, the Netherlands, Italy, Sweden, Norway, Poland, Hungary, and Israel. This period was characterized by extended periods of low economic growth and several economic recessions to which countries responded with quite different social policies. As a consequence, changes in mobility patterns over time and differences between countries were expected. These expectations were supported by the evidence (Breen and Luijkx 2004). The authors found a convergent trend across countries in total mobility. Countries differed in their level of relative mobility, with Germany, France, Italy, and Ireland showing less mobility than the rest. Finally, there was a widespread tendency toward greater relative mobility, except in the cases of Britain and Germany. In the present study, we extend the period of investigation covered by Erikson and Goldthorpe in the other direction. We study trends in the to2 The study of Ganzeboom, Luijkx, and Treiman (1989) also covers earlier birth cohorts than the study of Erikson and Goldthorpe (1992), but not in all countries. Since birth cohorts are not distinguished in Ganzeboom et al. (1989), it is not clear how this affected the results. 840 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility tal intergenerational class mobility of men and trends in relative mobility in seven countries (Britain, Finland, France, Germany, Hungary, the Netherlands, and Sweden) in the period 1800–1914. This is the first time that intergenerational mobility has been studied in a comparative way for this many countries and for such a long period. The historical extension of previous studies is essential because the main hypothesis on changes in intergenerational class mobility, also used in the studies by Ganzeboom et al. (1989), Erikson and Goldthorpe (1992), and Breen (2004), refers to the transition from preindustrial to industrial societies. This hypothesis can thus be properly tested only by reference to an earlier period. As a by-product, our analyses can also shed light on the question whether the trend toward more mobility in the second half of the 20th century, as observed by Ganzeboom et al. (1989), Breen (2004), and a range of country-specific studies, was the continuation of a trend starting during the Industrial Revolution. Alternatively, it has a bearing on the point of view of Erikson and Goldthorpe (1992) and Wong (1992) that a consistent trend over a broad range of countries did not take place, not recently and not in the 19th century. Why has the question on changes in intergenerational mobility during industrialization not yet been answered? Sociologists studying intergenerational mobility have mainly used survey data, and these data go back no further than the 1940s. Using retrospective data and birth cohorts, analyses can be extended to the beginning of the 20th century, but no further. In many developed countries the transition from a preindustrial to an industrial society was well underway by then. Historians, on the other hand, have used vital registers—that is, birth, marriage, and death registers— with occupational titles, which made it possible for researchers to track intergenerational change back further into the past. Until recently, though, gathering these data, sometimes linking data from several generations, and classifying occupational information from the sources in a comparable way, was very cumbersome and time consuming. Therefore, these studies were often restricted to small regions or compared only a few years (see the reviews by Kaelble [1981, 1985] and Miles and Vincent [1993]). This prevented firm conclusions being drawn about country-wide and long-term changes. The practical problems that inhibited the analysis of large amounts of historical data on intergenerational mobility have recently been resolved. In many European countries a sizable proportion of historical vital registers has now been digitized. A classification system geared to occupational titles from historical sources in different countries has been developed (van Leeuwen, Maas, and Miles 2002, 2004), along with a corresponding class scheme (van Leeuwen and Maas 2011). For the first time ever it is now 841 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology possible to answer the question whether total and relative mobility in European societies increased as they developed from preindustrial to industrial societies. HYPOTHESIZED EFFECTS OF INDUSTRIALIZATION ON INTERGENERATIONAL MOBILITY Total Mobility According to the “logic of industrialism” school, the total extent of intergenerational mobility in countries was lower before industrialization than after industrialization. Furthermore, proponents of this school argued that prior to industrialization the differences between countries were greater than afterward (Inkeles 1960; Kerr et al. 1973 [1960]; Form 1979). Industrialization was thought to have caused an increase in total mobility and a convergence of mobility patterns, for several reasons. During industrialization the size of certain classes, notably the farming class, decreased and that of other classes increased. As a consequence men were “forced” to leave the class of their father. Furthermore, managers in industrial societies recruited more on the basis of achievement instead of ascription in order to increase efficiency and decrease risks, thereby also decreasing the influence of a father’s class on that of his son. The attitude of workers to the world in general and labor market opportunities in particular also changed. Workers are said to have embraced modern universalistic values and to have responded by investing more in the education of their children and by grasping opportunities for upward mobility. Scholars from the logic of industrialism school expected a change in the extent of total mobility during industrialization, and that change would be both sizable and limited in duration. As soon as countries became industrialized, the mobility rate would remain stable at a high level. The reasoning followed by the logic of industrialism school is embodied in the classic Lipset and Zetterberg hypothesis (LZ hypothesis) of 1959, which states that social mobility became relatively high in societies once industrialization had reached a certain level there. A comparison of industrialized societies would show similar degrees of intergenerational mobility. Interestingly, Lipset and Zetterberg presented an overview of the determinants of change in mobility within industrialized societies: (1) changes in the number of vacancies; (2) different rates of fertility (of high- and lowstatus groups); (3) changes in the rank accorded to occupations; (4) changes in the number of inheritable status positions; and (5) changes in the legal restrictions pertaining to potential opportunities. However, the phrasing of their argument seems to imply that these determinants were either not 842 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility very important (for instance, with regard to changes in the rank of occupations), or similar in all industrialized countries, because, all told, Lipset and Zetterberg expected total mobility to be stable and similar in industrialized countries. Fukumoto and Grusky (1993) juxtaposed the threshold interpretation of the effect of industrialism with the view that class-based differences in mobility slowly withered away as universalistic values gradually spread and bureaucratic personnel systems were established. Among sociologists, this latter version of the industrialization thesis has found widespread currency (e.g., Blau and Duncan 1967; Treiman 1970), probably due to their focus on recent decades. For the period that we are studying, it is not possible to distinguish between the two interpretations of the industrialization hypothesis, because industrialization was still underway at the end of our period (1914). Reviews of historical studies conclude that total mobility did not in general increase during industrialization (Kaelble 1981, 1985; Schüren 1989). But it should be noted that these studies related to samples of small size and few points in time. Furthermore, the problem of noncomparability of results due to the absence of internationally comparable classification systems had not yet been resolved, thus leaving open a very wide margin of error. An interesting contrasting conclusion had earlier been drawn by Sorokin ([1929] 1959). Although his conclusion that “there seems to be no definite perpetual trend toward either an increase or a decrease of . . . mobility” is often cited (p. 152), he also concluded that “within Western societies during the last century there seems to have existed a trend toward a decrease of inheritance of occupation” (p. 458); the last century is, in this respect, the period 1829–1929. When comparing total intergenerational mobility between industrial societies, sociologists concluded that these societies differ greatly. In some countries the proportion of men reaching a class different from that of their father is much higher than in other countries (for reviews, see Ganzeboom, Treiman, and Ultee 1991; Fukumoto and Grusky 1993; van Leeuwen and Maas 2010). The lack of support for the LZ hypothesis in industrial societies led most sociologists to abandon the study of total mobility and to turn to studying relative mobility (e.g., Featherman, Jones, and Hauser 1975; but for exceptions see Simkus 1984 and Hout 1988). However, the conclusions of historical and sociological studies still leave us with the question where the observed contemporary country differences came from. Did similar country differences exist two centuries ago? Or were they even greater in preindustrial and industrializing societies, as the industrialization thesis predicts, and were historical studies just too small and too short term to uncover the changes? 843 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology Relative Mobility After having rejected the LZ hypothesis of cross-country similarity in total mobility in industrial societies, Featherman et al. (1975) formulated an alternative hypothesis on cross-country similarities in relative mobility in such societies. They stated that “circulation mobility” in industrial societies with a market economy and a nuclear family system is basically the same. This hypothesis differs from that of Lipset and Zetterberg not only in its focus on relative mobility but also because it restricts itself to Western types of industrial society and excludes societies lacking a market economy or a nuclear family system. Erikson and Goldthorpe (1987, p. 162) also pointed out a restriction to the generalizability of their findings: “A basic similarity will be found in patterns of social fluidity . . . across all nations with market economics and nuclear family systems where no sustained attempt has been made to use the power of the modern state apparatus in order to modify the processes, or the outcomes of the processes, through which class inequalities are produced and intergenerationally reproduced.” The similarity in relative mobility within and among Western industrialized societies over time is explained by the fact that in all these societies work activities are organized in basically the same way. As a result, occupational hierarchies show a large degree of cross-national commonality— which is reflected in broad similarities in both the desirability or prestige of different occupational groupings and their objective “socioeconomic” attributes, such as education and income (Featherman et al. 1975). In all industrial societies, thus, men strive to reach the same desirable class positions, but men from higher classes of origin have more resources to draw on in these attempts and are more likely to succeed. Featherman et al. do not explicitly compare industrial with nonindustrial societies. However, from what they write one can deduce that within the agricultural sector work is organized in a way different from that in the industrial sector. Possession of land is the most important resource in the agricultural sector and land is—as far as possible—kept within the family. As a result, preindustrial, agrarian societies are less open than industrial societies. Featherman et al. (1975) did not assume that industrial societies would become more open over time, but others did. According to Blau and Duncan (1967), this change was driven by a pervasive concern with efficiency in industrial societies. As a consequence, objective evaluation criteria became necessary, in the workplace as well as at school. A high position could no longer be directly inherited; it had to be legitimated by actual achievements that were socially acknowledged, that is, educational achievement. And in order to fill the increasing number of positions at the top of the occupational 844 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility hierarchy, higher levels of education were opened up to children from lower strata. Results of studies of relative mobility in past societies are even less conclusive than those of total mobility. Some studies found an increase in relative mobility during industrialization (e.g., Miles 1993; Lambert, Prandy, and Bottero 2007; Lippényi, Maas, and van Leeuwen 2013a;), but most did not (e.g., Fukumoto and Grusky 1993; Maas and van Leeuwen 2002).3 The latter findings, though, may arise from a lack of statistical power, given that many historical studies are small and cover only a brief period. A number of studies showed a slow increase in relative mobility within most industrialized societies (e.g., Ganzeboom et al. 1989; Breen and Luijkx 2004;), but some did not, including the influential study by Erikson and Goldthorpe (1992). As was the case with total mobility, there is more support for differences in relative mobility patterns between industrialized countries than for trends within them (Ganzeboom et al. 1989; Wong 1992). Very few mobility studies have compared countries over a long time period. Long and Ferrie (2013) have studied mobility patterns in the United States and Britain around 1880 and 1975, and Bourdieu, Ferrie, and Kesztenbaum (2009) compared the United States with France in the periods around 1880, 1910, and 1975. In both studies historical data (linked censuses and marriage registers) for the early period are compared with survey data for the second half of the 20th century. They concluded that around 1880 relative intergenerational mobility was higher in the United States than in Britain and France. The gap with Britain narrowed over time, mainly because relative mobility decreased in the United States, but that with France did not show a consistent trend. A decrease in relative mobility is an exceptional finding. Two studies reanalyzing Long and Ferrie’s data show that the decrease is driven by the specific long-term development of farmers’ intergenerational mobility that remains undiscovered in their summary measure of mobility (Hout and Guest 2013; Xie and Killewald 2013). If farmers’ immobility is taken into account, a decrease in mobility is no longer found. Again, the question remains regarding the origin of contemporary differences. Are they, as the industrialization hypothesis implies, the remnant of larger differences from the preindustrial period, fading away? Or have long-term changes been negligibly small and nonsystem3 The data used in this study overlap partly with those used in some of these previous studies. For England the data collated by Miles (1993) are used, but now with occupational coding and a class scheme comparable to those of the other countries. Nevertheless, we do not expect a different conclusion regarding the country-specific trend. The conclusions of Lippényi et al. (2013b) are based on the same Hungarian data, but for a longer (more modern) period. The studies of Maas and van Leeuwen (2002), van Leeuwen et al. (2016), and Bourdieu et al. (2009) are based (partly) on a small portion of the Swedish and French data, respectively. 845 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology TABLE 1 Predicted Effects of Industrialization on Total and Relative Mobility After the Onset of Industrialization 1. 2. 3. 4. 5. 6. Countries that were more industrialized showed higher rates of total mobility than countries that were less industrialized Total mobility increased Country differences in total mobility decreased Countries that were more industrialized showed higher rates of relative mobility than countries that were less industrialized Relative mobility increased Country differences in relative mobility decreased atic? Table 1 presents an overview of the hypotheses on the effects of industrialization. DATA AND METHODS Data The historical data sets used in this article cover 1.2 million marriages within seven European countries (Britain, Finland, France, Germany, Hungary, the Netherlands, and Sweden) in the period 1800–1914 (table 2). All data originate from vital registers that include records relating to births, deaths, and marriages. For present purposes, the marriage records are the most useful. Apart from other information, marriage registers contain the occupations of spouses and often those of their parents. For some countries (e.g., Britain and Hungary) vital registration records compiled by the church are used, for other countries marriage records compiled by the state (e.g., France and the Netherlands). Sometimes the marriage records are complemented by occupational information from death records, birth records, and population registers (e.g., Finish, Swedish, and some German data sets). In Finland, and in some German data sets, marriage records are linked across generations in order to create genealogies, or so-called clan registers. Whenever possible, only the first marriages of the bridegroom have been selected. In the case of Britain and of the Dutch Genlias data set, it was not possible to exclude higher-order marriages. Some data sets are comparable to surveys in the sense that they start from a (random, sometimes stratified) sample of the population. Examples include one of the Dutch data sets (the Historical Sample of the Netherlands), which starts with a random sample of Dutch birth registers, the Hungarian and British data sets, for which random samples of marriages are drawn from a stratified sample of cities and villages, and the French data set, 846 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Type of Data Coverage N Originala/ N Analyses Population registers Stratified sample of regions, complete population within regions b First marriages during the period included in the study. Because not only ancestors are included but also all the descendants of these ancestors, the final sample is fairly representative of the Finnish population for this period. a Sweden . . . . . . . 1800–99 Germany . . . . . . 1800–1914 Mixed See app. table A2 Hungary . . . . . . 1865–1914 Marriages (church) Stratified sample, present Hungary Netherlands . . . 1810–1914 Marriages (civil) See app. table A2 Miles (1999) Häkkinen (2006) Reference Dupâquier (2004); Leroy, Nicolas, and Pélissier (1998) First marriages 25,721/14,412 See appendix First marriages 36,939/31,060 Lippényi, Maas, and van Leeuwen (2013a) Differs between 1,051,608/517,420 See appendix data sets First marriages 67,908/23,384 CEDAR 2015 Britain . . . . . . . . 1835–1914 Marriages (church) Stratified sample, England All marriages 10,663/9,836 Finland . . . . . . . 1800–1914 Genealogies Ancestors of 10 poor families in First marriages 6,185/4,257 1930 and all their descendantsb France . . . . . . . . 1800–1914 Marriages (civil) Name sample, France First marriages 44,913/24,452 Period TABLE 2 Characteristics of the Data Sets American Journal of Sociology which consists of all vital registers of persons whose surname starts with the letters “Tra” (these letters were chosen because surnames starting with these three letters occur in all languages spoken in France and are equally distributed over social classes). Other data sets cover in principle all marriages that took place in a certain region. In the case of one of the Dutch data sets (Genlias), this region covers almost half of the country, resulting in over a million cases. Other regional data sets, from Germany and Sweden, are smaller. With the exception of Germany, the coverage of all data sets is so large that they can, more or less, be considered to represent the country as a whole. For Germany, no such data set exists. To be able to include Germany in the analyses, we use six smaller German studies (app. table A1). Not all data sets cover the whole period from 1800 to 1914. This is sometimes the result of the sources themselves—in most cases civil marriage registration did not commence until around 1812, and Hungarian data contained very little occupational information before 1865. Sometimes this lack of completeness was a deliberate decision on the part of the data collectors, or it was driven by privacy regulations. For example, the Swedish data cover exactly the 19th century, and the data on Berlin are from just certain years between 1825 and 1909. Although a few data sets continue after 1914, we decided not to include those years because of the possible major influence of the First World War in some of the countries being studied. Whereas the main problem of survey research is selective nonresponse, the main problem of the historical data sources used in this study is selective missing information on the occupation of the father. The number of cases that can be used in the analyses is only about half of the original number of cases. This difference is caused mainly by the fact that the marriage certificate did not contain information on the father’s occupation, usually because the father was not present at the marriage of the son, most often because he was no longer alive or sometimes because the son had migrated over a large distance. It is often assumed that sons who migrated or whose father died when they were still young might have been less affected by the class of their father, causing us to underestimate mobility. Previous studies using the French and the Dutch data sets, however, report very little selective dropout (Zijdeman 2009; Maas et al. 2011). In table 3 we compare, for all the countries, the class distribution of sons before and after deleting cases without information on the occupational class of their father. If unskilled workers were more likely than other classes to migrate over long distances, we should see, after deleting cases without information on father’s class, a smaller percentage of sons from this class in table 3. Higher mortality of fathers from the unskilled working class would have the same result, under the assumption that there is an association between father’s and son’s 848 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 11.0 25.4 4.3 55.7 3.6 7.8 10.9 25.4 4.5 55.5 3.7 A 7.5 5.9 48.9 3.5 34.2 B 31.2 6.9 4.9 52.1 2.7 33.3 A Finland 13.2 19.4 34.4 26.5 6.6 B 45.6 A 12.6 19.8 35.8 25.4 6.3 France 13.1 22.8 20.0 38.4 5.7 B 44.0 12.3 23.8 17.9 40.9 5.2 A Germany 12.5 18.1 39.8 14.5 15.1 B 16.0 11.0 17.9 41.0 14.5 15.5 A Hungary 13.3 16.3 19.2 37.6 13.6 B 50.8 12.5 17.0 18.6 37.8 14.1 A Netherlands NOTE.—Germany and the Netherlands: results pertain to all data sets combined. For all countries, B 5 before; A 5 after. Nonmanual . . . . . . . . . Skilled workers . . . . . . Farmers. . . . . . . . . . . . Unskilled workers. . . . Farm workers . . . . . . . % missing . . . . . . . . . . B Britain TABLE 3 Class Distribution of Sons before and after Deleting Cases without Information on Occupational Class of Father 8.3 7.7 21.0 20.5 42.4 B 65.6 4.7 4.9 31.8 13.6 45.0 A Sweden American Journal of Sociology classes even after the death of the father (an assumption that is supported in empirical research; see van Poppel, de Jong, and Liefbroer 1998). Our results confirm the conclusions about the lack of selective dropout in the French and Dutch data sets, and this appears also to be the case for the British, German, Hungarian, and Finnish data.4 Only the data set from Sweden shows an overrepresentation of sons who are farmers—and thus unlikely to be migrants. This is a consequence of the regional character of the Swedish data in combination with the fact that the occupational information comes from the population registers. A father’s occupation is not given on his son’s marriage certificate, but if the father lived in the same region, his occupation can be found in the population register. If the son had migrated from the region where his family of origin lived, his father’s information would be lacking. This means that Sweden was probably somewhat less agrarian and more mobile than what is reflected in the data. We further investigate the absence of father’s occupation from son’s marriage certificate with data from the Historical Sample of the Netherlands (HSN; Release Life Courses 2010.01), an exceptional data set that includes both marriage acts and population data of the same persons. These data cover the period after the onset of industrialization (1888–1939). Men for whom father’s occupation was missing on the marriage certificate but could be retrieved from the population register were more often intergenerationally mobile than men for whom father’s occupation was recorded on the marriage certificate.5 Absolute mobility was, respectively, 48.9% versus 4 Note that in all countries the percentage of sons in the highest class is smaller after deleting cases without information on occupational class of the father. This may indicate that sons from this class more often migrated over a long distance. It speaks against the idea that sons of deceased fathers would be more often downwardly mobile. 5 We started with analyzing all marriage certificates from the HSN marriage cohorts 1888–1939 that have information on bridegroom’s occupation (N 5 8,521). Father’s occupation is missing in 43.1% of these cases (valid N 5 4,852). We compared the results based on these data with those for all cases in which we have information on father’s occupation from either the marriage certificate or the population registers (father’s occupation now missing in only 14.1% of all cases; valid N 5 7,319). Sons from fathers for whom no information was available on the marriage act, but for whom we did find information in the population registers, were more often intergenerationally mobile than sons from fathers with information on the marriage act (48.9% vs. 39.6%). Relative mobility was also higher (uniform difference parameter 0.73). In a second step we compared sons of fathers with information on the marriage act (the type of population that we analyze in our study) and the “total population”—in this case sons of fathers with information from marriage acts or population registers, distinguishing between early marriage cohorts (1888–1908) and late marriage cohorts (1909–39). In the total population the increase in total and relative mobility is somewhat smaller than in the marriage certificate population (total mobility: 5.1% vs. 7.2% increase; relative mobility: uniform difference parameter 0.91 vs. 0.88). However, the change is in the same direction and highly significant in either sample. Thus, although sons whose father very likely died before the son married were more mobile than sons whose father survived, the main conclusion on 850 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility 39.6%, and relative mobility was significantly higher as well. This is in line with the results of van Poppel et al. (1998). Thus, sons of fathers who had died relatively young and sons who had migrated over a large distance were indeed more mobile than other sons. As a consequence, observed mobility from marriage records only is lower than that of the “total population” (as captured via marriage and population data combined), at least in this Dutch case. However, the trend remains the same: when we compare early marriage cohorts (1888–1908) with late marriage cohorts (1909–39) mobility is always higher in the latter cohorts, whether we look at marriage acts or at the total population. Although this is a reassuring finding, it remains the case that mobility in the two populations was different. Since data as rich as those collected in the HSN are very rare, we cannot investigate whether similar differences existed in the other countries. If greater mobility among men with marriage certificates on which the occupation of the father is not recorded was generally the case, one would expect to find a negative relationship between the percentage of cases without information on father’s occupation and the amount of mobility. As a robustness check we will therefore investigate whether our conclusions hold true or not, when we take the percentage of missing cases into account. A specific problem associated with marriage data is, of course, that these cover only the population marrying. Nowadays, marriage data are therefore far from representative for the population. However, in the period covered by this study the percentage of men still unmarried at age 45–49 ranged between 13 and 11 in France, 12 and 10 in Britain, 11 and 13 in the Netherlands, and 7 and 10 in Sweden (Hajnal 1953); the percentage was even lower in Hungary (Hajnal 1965). Changes over time were small, and remaining single was in general not more or less likely after the onset of industrialization than before. It could, however, be argued that downwardly mobile men were not only less successful on the labor market but on the marriage market as well and thus more likely to stay single. And if this relationship between marital status and mobility changed over time, our results might be biased. This issue can also be investigated using the HSN data (2010), which include both married and single men. Married men were indeed more likely to be upwardly mobile than single men (20.3% vs. 13.6%). The difference increased over time (from 3.2% for men who married between 1888 and 1908 to 8.5% for those who married between 1909 and 1939). However, due to the fact that singles were a minority and the differences in mobility are relatively small, an analysis of the total (married and single) male population leads to very similar results and to the same substantive conclusions as an analysis of married men only. Total trends in mobility remained the same, irrespective whether we analyzed the marriage certificates or the total population. 851 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology mobility numbers never differed more than 1% between the total and married populations, and the difference in the relative mobility parameters was negligible.6 Since in the other countries we studied singlehood is not more common than in the Netherlands and there are no obvious reasons to expect larger differences in mobility between singles and married men, we are confident that population-wide trends will closely follow trends among married men. In some respects, marriage data are more suitable than survey data for studying trends in mobility. In the intergenerational mobility tables based on the occupations at marriage, the class position of sons at their own marriage is compared with that of their father at the marriage of their son. The son’s occupation is thus measured at more or less comparable ages (around age 27).7 It is likely that men would have married after they had attained a more or less stable occupation that allowed family formation. In survey data the occupation of the sons is usually measured at very different ages and thus different stages in the occupational career. Of course, it is possible to restrict survey analyses to sons in a certain age range, but this is not often done, maybe because it involves a considerable decrease in the number of cases (e.g., Ganzeboom et al. [1989] analyzed men ages 21–64 and Breen and Luijkx [2004] men ages 25–64). Furthermore, marriage data are better than survey data at allowing one to pinpoint intergenerational mobility in historical time. Whereas first occupations might take the form of small, temporary jobs, relatively unrelated to a father’s occupation, an occupation that allowed family formation is likely to have been both strongly dependent on parental resources and a very strong determinant of a son’s further We compared all married men (N 5 8,653) with all men who remained single and who are observed at least until age 40 (N 5 888). In a first step we investigated whether singles are from lower-status background than married men. That appeared not to be the case. The only significant difference is found for sons of farmers who were more likely to remain single (14.8%) than sons from other classes (between 7.7% and 10.4%). In a second step we compared absolute mobility of single and married men. Among married men 17.4% were downwardly mobile, and among single men, 16.7%, which is not significantly different. But married men were more likely to be upwardly mobile than single men (20.3% vs. 13.6%), and this difference increased over time (from 3.2% for the marriage cohort 1888–1908 to 8.5% for the marriage cohort 1909–39). In a third step we investigated the effect of this changing difference in mobility between married and single men on total and relative mobility by comparing mobility tables based on married men only and based on the total male population. The total mobility for the first cohort was 34.7% (married men) and 34.1% (total population) and for the second cohort 39.1% and 38.1%. The uniform difference parameter for the second cohort compared to the first was 0.86 for married men and 0.87 for the total population. We conclude that the effect of excluding single men on the mobility parameters is negligible and excluding single men does not lead to different conclusions. 7 Mean age at marriage differed little between countries, ranging from 26.9 (Britain) to 28.1 (Finland). It changed only slightly over time. Differences over time in mean age within one country never exceeded four years. 6 852 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility occupational career (Blossfeld 1987; Schulz 2013). At marriage, or just before marriage, important processes causing intergenerational mobility, or a lack of it, thus took place. Since the year of marriage is known, these family processes can be linked to what occurred in society at large at that time. Researchers analyzing survey data may use birth cohorts to approach the important early stage of the occupational career (e.g., Erikson and Goldthorpe 1992), but they would usually lack a measurement of occupation at that time. Or they might assume that historical processes were at work at the time the son’s occupation was measured, though that could be at very different stages in these sons’ occupational careers (e.g., Ganzeboom et al. 1989; Breen and Luijkx 2004). A last attractive characteristic of marriage data is that in principle these are available for each year, allowing precise modeling of change over time. Measures All occupational information relating to fathers and sons has been coded uniformly using the Historical International Standard Classification of Occupations (HISCO), an occupational classification system that is both international and historical (van Leeuwen et al. 2002, 2004). Standard routines are available to transform HISCO codes into the historical class scheme HISCLASS (van Leeuwen and Maas 2011). HISCLASS distinguishes 12 classes: (1) higher managers, (2) higher professionals, (3) lower managers, (4) lower professionals and clerical and sales personnel, (5) lower clerical and sales personnel, (6) foremen, (7) medium skilled workers, (8) farmers and fishermen, (9) lower-skilled workers, (10) lower-skilled farm workers, (11) unskilled workers, and (12) unskilled farm workers. Because some classes are only sparsely populated, we combine classes 1 to 5 into a nonmanual class, classes 6 and 7 into skilled workers, classes 9 and 11 into unskilled workers, and classes 10 and 12 into farm workers, leaving a total of five classes. Using HISCO and HISCLASS it was possible to classify sons and fathers comparably across time and space. We analyze intergenerational mobility in our seven countries against the background of industrialization. In order to describe industrialization processes in the countries included in this study we use two indicators: the percentage of sons in the agricultural classes (combining farmers and farm workers) and the percentage of sons in industrial occupations. Industrial occupations include all HISCO categories that involve working with a machine or electrical device. These occupations range from telegraphers and bus conductors to machinery mechanics and crane drivers. We were conservative in assigning an occupation to the industrial sector; occupations that can also be performed by hand are not included. For example, in the broader HISCO group of weavers and related workers the following occu853 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology pations are considered industrial: beam warper, loom threader (machine), cloth weaver (machine, except Jacquard loom), lace weaver (machine), carpet weaver (machine), and net maker (machine). Other weavers such as weaver, specialization unknown and cloth weaver (hand) are considered nonindustrial because it is either unclear whether a machine is involved, or it is explicitly stated that this is not the case. Nonindustrial occupations that grew in size as a consequence of industrialization, for example, those of secretaries and bank employees, are not counted as industrial either. It is likely that a considerable proportion of those who worked in the industrial sector have not been captured by our conservative definition. This is because bridegrooms whose occupational titles leave it uncertain that they use a machine end up in unspecified HISCO categories (e.g., weaver, specialization unknown). Methods In order to calculate total mobility percentages and relative mobility parameters, mobility tables were created for each five-year period, starting with 1800–1804 and ending with 1910–14. If the number of cases in a table was fewer than 300, the table for this period was combined with that of an adjacent period. In total 16 tables are analyzed for Britain, 10 for Finland, 22 for France, 36 for Germany (from six data sets), 10 for Hungary, 36 for the Netherlands (from two data sets), and 19 for Sweden. The total degree of mobility is measured as the percentage of sons in a class that is different from their father’s class. In order to investigate whether total mobility increased more after the onset of industrialization than before, logistic regression models are estimated including separate time trends (splines) for years before and after the onset of industrialization. A log-linear model is estimated to compare relative mobility between countries and periods, because it excludes effects of the marginal distributions of the mobility tables, for example, caused by changes in the occupational distribution. What is left is called relative mobility: the relative chances of sons from two different classes of origin ending up in one specific class and not another (comparable to odds ratios). We estimate a uniform difference model, which is especially suited to investigating differences in relative mobility between mobility tables. It does not restrict the pattern of mobility within a table but summarizes the difference in relative mobility between tables in one multiplicative—uniform difference—parameter (Erikson and Goldthorpe 1992; Xie 1992). We also estimate models with two multiplicative parameters per table: one for the overall pattern of relative mobility and one for additional immobility on the diagonal. With these models we explore whether differences between countries and periods mainly exist with respect to immobility or with respect to the class in which 854 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility men end up if they leave their class of origin (i.e., when they are intergenerationally mobile). The models are estimated using lEM (Vermunt 1997). The fit of the models is compared using a v2 test and the Baysian information criterion (BIC). In a second step we will test our hypotheses on country differences and trends in relative mobility by analyzing the uniform difference parameters with a linear regression model, again including splines. We decided not to weight the mobility parameters by a measure of their reliability, because that would cause the results to be driven mainly by the very large Dutch data set. Instead, we analyze unweighted parameters. We compare country-specific analyses with the overall analyses in order to draw conclusions on the generalizability of our results. Hypotheses on the convergence of the countries with respect to intergenerational mobility are tested using coefficients of variation. We calculate coefficients for different combinations of countries for several reasons. Not all countries cover the whole period, thus the size of one overall coefficient would be driven by countries entering and leaving the set of observations. Therefore, we select groups of countries with similar observation periods. Further, we investigate whether results are stable if we include Germany—the country with the least representative data. The data sets differ from each other in some respects. In the Swedish data set the lack of occupational information for a considerable number of fathers led to an overrepresentation of farming bridegrooms. The British data and the Dutch Genlias data set include not only first but also second and higher-order marriages. We perform a number of robustness checks to investigate whether differences between data sets affect our results. We compare results of analyses with and without higher-order marriages. For Sweden we compare analyses with and without weighting to correct for the overrepresentation of farmers. And for the Netherlands we compare the results based on the Genlias data set with those based on the HSN data set. Finally, we investigate whether our results with respect to relative mobility still hold when the specific immobility of farmers is taken into account by adding a specific parameter for this cell in the mobility table that can vary freely over time and country (compare Hout and Guest 2013; Xie and Killewald 2013). RESULTS Industrialization Britain was the first country to industrialize, as early as the 18th century, and this is clearly visible in figures 1 and 2. Figure 1 illustrates that the proportion of bridegrooms working in agriculture in England was around 0.10 855 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology FIG. 1.—Indicators of industrialization in seven European countries, 1800–1914: proportion in agriculture. The Dutch analyses pertain to the Genlias data and the German analyses to the Six Villages data. over the whole period observed, a much lower proportion than in the other countries. Figure 2 illustrates that the proportion of bridegrooms in industrial occupations started to increase earlier in Britain (around 1850) than in the other countries. Germany, France, and the Netherlands formed a middle group with respect to industrialization. At the beginning of the 19th century, 40%–50% of bridegrooms worked in agriculture, and this percentage decreased only slowly. The proportion of bridegrooms working in industrial occupations started to grow 20 years later in the Netherlands, Germany, and France (around 1870) than in Britain. As one might expect, German regions and cities differed strongly with respect to industrialization. In none of the regions, though, did the percentage of bridegrooms working in agriculture FIG. 2.—Indicators of industrialization in seven European countries, 1800–1914: proportion in industrial occupations. The Dutch analyses pertain to the Genlias data and the German analyses to the Six Villages data. The proportions are conservative estimates. 856 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility change much over time. In the cities of Berlin, Bielefeld, and Borghorst the proportion of bridegrooms working in industrial occupations started to increase earlier than in the Six Villages data set. The two Dutch data sets show very similar developments in terms of the proportion of bridegrooms working in agriculture or in industrial occupations. The Genlias data cover six provinces in the east of the Netherlands. The largest cities in the Netherlands are in the west of the country, which is reflected in a somewhat higher percentage of bridegrooms in agricultural occupations in the Genlias data set compared with those in the nationwide HSN sample.8 Sweden, Finland, and Hungary industrialized late, with very high proportions of men working in the agricultural sector throughout the 19th century.9 But especially in Sweden, the proportion working in agriculture decreased. The proportion of bridegrooms in industrial occupations began to grow around 1890. To what extent are the indicators of industrialization from our own data consistent with what we know from the literature about industrialization in Europe? An indicator of industrialization often used is innate energy, that is, not from human labor or the use of animals (Davis 1955). In its operationalization this concept is often captured by looking at steam, the earliest main source of modern energy (disregarding energy from wind and watermills). From surveys of the economic history of Europe (e.g., Persson 2010; Broadberry and O’Rourke 2010a, 2010b; Berend 2013) it is clear that steam engines were first used in Britain, then spread to Belgium, France, and Germany, and later to much of the rest of Europe, much later in fact in areas further away from the epicenter such as Russia and present-day Hungary (which, with Austria, formed the Habsburg Empire). Comparative measures of the use of steam are rare, however. A minor exception is formed by data assembled and published by the French National Bureau of Statistics (Statistique Générale de la France 1926) for four countries (Britain, France, Germany, and Sweden) in the years 1905–7 (see table 4). Even at this late date, the trifold division into core (Britain), semiperiphery (here covered by France and Germany), and periphery (here Sweden) suggested by the literature is evident. It should be noted here that historical descriptions of industrialization suggest that France industrialized earlier than Germany, which caught up, however, from the late 19th century. If a similar table could be presented for an earlier period, the gap between France and Germany would be smaller, and, going back to the early 19th century, the position of the two countries would be reversed. 8 Figures for the separate German and Dutch data sets are available upon request. The indicators of industrialization have been calculated with respect to all bridegrooms, irrespective of whether their father’s class was known. The Swedish figure thus suffers less from selectivity due to missing values. 9 857 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology TABLE 4 Indicators of Industrialization in European Countries, 1800–1913 COUNTRY Britain . . . . . . . . . . . . . . Finland . . . . . . . . . . . . . France . . . . . . . . . . . . . . Germany . . . . . . . . . . . . Hungary (1Austria) . . . Netherlands . . . . . . . . . Sweden . . . . . . . . . . . . . HORSEPOWER 1906a 9,765 2,605 6,715 232 PER CAPITA LEVELS OF INDUSTRIALIZATIONb 1800 1830 1860 1880 1900 1913 16 8 9 8 7 9 8 25 8 12 9 7 9 9 64 11 20 15 11 11 15 87 15 28 25 15 14 24 100 18 39 52 23 22 41 115 21 59 85 32 28 87 a The horsepower of steam engines multiplied by 1,000, with data taken from Statistique Générale de la France 1926:310–11. b For per capita levels of industrialization, the years are 1800–1913 for the countries in the sample (UK 1900 5 100); Bairoch (1982), pp. 281, 294, and 330. A second often-used indicator (Bairoch 1982) refers to the size of the secondary sector (industry) in the labor force (consisting, in addition, of those working in the primary sector—agriculture—and the tertiary sector—services). In this sense industry is much broader than the concept of mechanized industrial jobs that we would ideally like to capture. Artisans such as coopers, blacksmiths, or tanners using age-old technology also worked in the secondary sector.10 These data do not contradict the notion of a starter (Britain), followed first by France and Germany, and later by Finland, and Hungary (table 4). For the Netherlands, the published figures— indicating late industrialization—deviate from our own analyses, in which this country takes a middle position. We assume this can be explained by the fact that the Netherlands already had a large service sector in the 19th century (Grafe, Neal, and Unger 2010, p. 212). As a consequence the relative size of the total secondary sector was small. Within that sector, though, increasingly traditional ways of production were replaced by industrial methods. Sweden also has a large secondary sector, but based on its very low score on horsepower in 1906 and the lack of a sharp increase in the share of bridegrooms in industrial occupations until the end of the 19th century, we assume that a large share of this secondary sector consisted of artisans. 10 Furthermore, these comparative data go back to estimates by Bairoch (1982), the derivation of which is not entirely transparent, and with the exception of data for Britain in 1800 (Broadberry, Fremdling, and Solar 2010, p. 172; based on Crafts and Harley 1992), they have not been revised. It is believed, however, that these data “fit well with the large secondary literature on the subject, and can at least be seen as providing a broad guide to the orders of magnitude” (Broadberry et al. 2010, p. 171). 858 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility An as yet unaddressed issue is how to dichotomize the era we are studying into a less and a more industrialized period. For the sake of simplicity we will refer to these periods as “preindustrial” and “industrial,” although “pre- and early” versus “mature” industrialization would be more sophisticated labels. We will consider 1850, 1870, and 1890 as the turning points, that is, for Britain (1850), for France, Germany, and the Netherlands (1870), and for Finland, Hungary, and Sweden (1890), because of the sharp increase in the proportion of industrial occupations among young men in these countries around those years. This periodization is made for the sake of the analyses. In fact it dichotomizes a continuous process of the spread of steam engines from Britain to the rest of Europe. Total Mobility As discussed above, total mobility is expected to have changed during industrialization through two processes. First, changes in the occupational distribution “forced” men to leave the class of their father. Second, managers in industrial societies recruited more on the basis of achievement instead of ascription, thereby decreasing the possibilities of fathers to transfer their class position to their son. Mobility resulting from the first process is often called structural mobility and that resulting from the second process, relative mobility. Figures 1 and 2 illustrated that the expected changes in the class distribution took place in all countries in our study sometime between 1800 and 1914. A simple measure summarizing all differences in the class distribution of fathers and sons is the index of dissimilarity.11 We calculated this measure for all mobility tables. It correlates .45 with the amount of total mobility, indicating that total mobility is indeed partly, but definitely not completely, driven by changes in the class distribution. In the following we analyze whether total mobility was higher in countries that industrialized earlier and whether it increased with industrialization. Figure 3 shows that the likelihood of intergenerational mobility differed considerably between countries. Whereas in Hungary and Finland at the time of their marriage between 20% and 30% of sons were in a class different from that of their fathers, in Germany this figure was between 40% and 50%. In Britain, the country that industrialized earliest, total mobility was no higher than in countries that industrialized later (France, e.g.) or much 11 The index of dissimilarity is defined as 1 N bi wi 2 o 2 i51 B W in which bi is the number of fathers in class i, B is the total number of fathers, wi is the number of sons in class i, and W is the total number of sons. 859 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology FIG. 3.—Total mobility (proportions) in seven European countries, 1800–1914. The line for Germany is predicted using model 2 in table 5. later (Sweden, e.g.). Germany and Sweden show the highest degree of mobility. Our first hypothesis, namely, that more industrialized countries can be expected to have greater overall mobility than less industrialized countries, is thus not supported by the historical data. A closer look at the country-specific changes over the whole period shows that in the majority of our countries intergenerational mobility increased over time (figure 3 and table 5, model 1). In Britain, with every 10 years the natural logarithm of the odds of being intergenerationally mobile increased by 0.09, which means the odds increased 9% every 10 years (e.09). France, Hungary, and Finland show an increase as well. The increase was smallest in Finland (3% every 10 years). No overall trend toward greater total mobility is observed in the Netherlands (a 2% decrease every 10 years), Sweden (a 1% decrease every 10 years), and Germany (no change). 860 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility TABLE 5 Changes in Total Mobility over Time Britain Finland France Germany Hungary Netherlands Sweden Model 1: Constant . . . . . . 2.97** 21.03** 2.04** Time . . . . . . . . . .09** .03* .04** Model v2 (df ). . . 87.83 (1) 5.92 (1) 85.99 (1) Model 2:a Constant . . . . . . 2.87** 2.98** 2.79** Preindustrial time . . . . . . . . 2.04 .02 .06** Industrial time . . . . . . . . .09** .17 .02 Model v2 (df ). . . 89.22 (2) 8.42 (2) 91.08 (2) N............ 9,836 4,257 24,452 2.12 .00 254.77 (6) 21.29** .08** 56.84 (1) 2.41** 2.02** 531.60 (1) 2.33** 2.01* 5.80 (1) 2.04 2.93** 2.25** 2.28** 2.03* 2.15** 2.08** 2.03** .06** .23** .03** .47** 267.25 (7) 1,552.31 (2) 1,231.06 (2) 30.89 (2) 14,412 31,060 509,344 23,384 NOTE.—Coefficients (b) of logistic regression of intergenerational mobility on time, using splines. The Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data set being the reference category). a Model 2 contains splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. According to the industrialization hypotheses, however, total mobility is expected to increase especially after industrialization. We therefore estimate a second set of models to investigate whether total mobility increased at a faster pace after industrialization than before, as theory predicts. In five of the seven countries, Britain, Germany, Hungary, the Netherlands, and Sweden, this was indeed the case (table 5, model 2). In Britain the overall increase in total mobility over time—which is visible in model 1—can be decomposed into a nonsignificant change during the preindustrial period and a significant increase of 10% every decade after the onset of industrialization (table 5, model 2). In Germany, Hungary, the Netherlands, and Sweden the likelihood of being intergenerationally mobile actually decreased in the preindustrial period, but it certainly increased thereafter (the odds rising 6%, 26%, 3%, and 60%, respectively, per decade). Finland and France both deviated from the general pattern of a clear increase in total mobility after industrialization. France showed rising mobility in the preindustrial period but not in the industrial period. In Finland changes in total mobility were not significant. We performed a number of robustness checks in order to investigate whether the results reflect a particular data set (Germany and the Nether861 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology lands), the inclusion of higher-order marriages (Finland, France, and Hungary), or the selective dropout of migrants (Sweden). For Germany the degree of mobility differs between data sets (app. fig. A1). Mobility was higher in Berlin and the Six Villages data set than in Ost-Friesland and the Hamburg and Villages data set. This was to be expected because the data sets are relatively small and differ greatly with respect to urbanization. It is reassuring that all the German data sets, as far as observed, show relatively high total mobility in the later decades of the 19th century. But it is unclear whether Ost-Friesland, which showed total mobility decreasing until 1870, would have later “joined” the other regions. The lack of an overall trend in total mobility in Germany—despite the separate figures suggesting otherwise—indicates that the inclusion and exclusion of data sets affects the results and that they are probably less reliable than those for the other countries. The two large data sets from the Netherlands show very similar degrees of and trends in total mobility, indicating that either of them can be used to describe the whole country. Whether or not higher-order marriages are included in Finland, France, Germany, or Hungary does not matter for the results (app. table A2). This gives confidence in the results for Britain and for the Dutch Genlias data, in which higher-order marriages cannot be distinguished from first marriages. Finally, for Sweden we compare weighted and nonweighted data. The weighted data represent the class distribution of sons before cases without information on the father’s class were deleted. The weighted data contain fewer farmers and consequently show more intergenerational mobility. However, the overall shape of the two Swedish lines in figure A1 is rather similar. The logistic regression models estimated on the weighted data show that the decrease in the preindustrial period is not significant, but the increase after the onset of industrialization is. Whether the results are affected by country differences and changes over time in the percentage of cases without information on father’s class is analyzed by adding this percentage as a variable to the models (app. table A3). The effect of the control variable is not significant in Britain, France, and Sweden, is negative in Finland and the Netherlands, and is positive in Germany and Hungary, possibly reflecting differences in the causes of information being missing between countries. In general, however, the conclusions are not affected by including the control variable. There is one exception. Whereas Finland did not show an increase in absolute mobility over time in the industrial period in our main analyses, it does if we take the percentage of cases with missing information on father’s occupation into account. This strengthens the conclusion that social mobility increases faster after industrialization than before. The sign and significance of all other effects do not change, with the exception of the trend in Germany in the preindustrial period. This trend was negative in our main analyses but is not signif862 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility icant in the analyses with the control variable, which is still in accordance with our conclusion. One would be justified in concluding that our second hypothesis—that total mobility rose during industrialization—is supported in most countries. According to our third hypothesis, country differences in total mobility decreased during industrialization. Coefficients of variation are calculated to test this hypothesis. The results are presented in figure 4. Since countries cover different periods, we present results for different combinations of countries. Whatever combination we use, however, the hypothesis is never supported. The countries converged with respect to their total mobility between 1800 and 1870, a period that, for all countries except Britain, has been considered preindustrial. After 1870 the countries start to diverge again. How much depends on the subset of countries. The variation in the preindustrial period is smaller and total mobility converged less in this period if we exclude Finland. Similarly, part of the divergence during industrialization is the result of Hungary entering the data set. The drop in variation in the last decade is caused by the large increase in total mobility in Hungary during this period, causing Hungary to increasingly resemble other countries. This detail in the graph suggests that the divergence observed after the onset of industrialization might have been temporary (fig. 4). The overall conclusion, however, is that industrialization definitely did not FIG. 4.—Convergence and divergence of total mobility in seven European countries, 1800–1914 (coefficients of variation). 863 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology cause an immediate convergence in total mobility and that there is actually more support for country differences having increased. Relative Mobility Changes in class structure, as reflected in the marginal distributions of the mobility tables, were important forces behind changes in total mobility. We have used log-linear models to analyze relative mobility, that is, the mobility that remains after taking into account the effects of the marginal distributions of the mobility tables. We first estimated a uniform difference model that allows the strength of the association to be different between all mobility tables. This model fits the data significantly better than a model that requires the association to be the same in all tables (D v2 5 6,015, D df 5 128 / BIC 218,938 vs. 216,550).12 Figure 5 shows the parameter estimates of the uniform differences in association between the tables. Britain in the first observation period (1835–39) is the reference point (fixed at 1). All other values are estimated multiplicative parameters that indicate how much stronger or weaker the association in a mobility table is compared with Britain 1835–39. We investigate whether country differences and changes over time are significant by analyzing the uniform difference parameters in a linear regression analysis. Figure 5 and table 6 show both significant country differences and historical changes. Germany, Sweden, and Finland were relatively open countries in which the association between father’s and son’s class was, on average, 0.39, 0.31, and 0.22 lower, respectively, than in Britain (model 2).13 France is more similar to Britain, but still somewhat more open (0.09 lower). The Netherlands and Hungary were relatively closed societies, with an association 0.07 and 0.09 higher, respectively, than in Britain. Thus country differences in relative mobility generally do not reflect country differences in the level of industrialization during the 19th century. 12 We do not want this overall test to be affected by differences between the German data sets. Therefore the model includes five parameters for the differences between the six German data sets. The association is thus allowed to differ between German data sets, but not between periods within the same data set or between Germany and the other countries. 13 The difference of 0.35 between Germany and Britain means that the natural logarithms of the odds ratios in the mobility table for Germany are on average 35% smaller than those in the mobility table for Britain. As such it is quite an abstract number since it does not show the actual strength of the odds ratios or the pattern of odds ratios within the tables, let alone the spread of cases over the cells of the mobility table. However, it is a convenient and widely accepted way of summarizing the difference in the relative chances of sons from different classes of origin to reach different classes of destination (i.e., relative mobility). 864 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility FIG. 5.—Relative mobility (uniform difference in association) in seven European countries, 1800–1914. The line for Germany is predicted using model 2 in table 7. Britain (1835–39) is the reference point (fixed at 1). As a first robustness check, we investigate whether this conclusion is affected by country differences in the percentage of cases with missing information on father’s class (app. table A4, model 2a). Country differences only slightly change in these analyses. Germany, Sweden, and Finland were still the most open societies, although Sweden moves closer toward Britain. France was still similar to Britain, and Hungary and the Netherlands were the most closed societies. Second, we analyze whether the trends and country differences are driven by farmers’ immobility (app. table A4, model 2b). Country differences in relative mobility indeed seem partly driven by farmers’ immobility. If we filter out farmers’ immobility, Hungary and the Netherlands were no longer especially closed societies. They become part of a group of countries with similar amounts of relative mobility, to which also Britain and France belonged. Finland and Germany were still relatively open societies, but the differences with Britain are smaller when farmers’ immobility is excluded than when it is included. Sweden is the exception 865 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology TABLE 6 Changes in Relative Mobility over Time and in All Countries Constant . . . . . . . . . . . . . . Time . . . . . . . . . . . . . . . . . Preindustrial time . . . . . . . Industrial time . . . . . . . . . Britain . . . . . . . . . . . . . . . . Finland . . . . . . . . . . . . . . . France . . . . . . . . . . . . . . . . Germany . . . . . . . . . . . . . . Hungary . . . . . . . . . . . . . . Netherlands . . . . . . . . . . . Sweden . . . . . . . . . . . . . . . Adjusted R2 . . . . . . . . . . . . Model 1 Model 2 .99* 2.00 1.03** .01* 2.03** 2.19** 2.06* 2.35** .15* .10** 2.28** .79 2.22** 2.09** 2.39** .09** .07* 2.31** .82 NOTE.—Unstandardized coefficients of regression analysis of uniform difference parameters on country and time, using splines. The Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). N 5 134. a Model 2 contains splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. to the general pattern. Without farmers’ immobility, the openness of Sweden is even more exceptional than with farmers included. We conclude that our fourth hypothesis, on country differences, is not supported. The late industrializing countries Sweden and Finland were more open than expected, and Britain, the most industrialized country, did not show especially high levels of relative mobility. From country differences we now turn to changes in relative mobility over time. There was no overall trend toward more openness during the whole period observed. However, if we consider trends in the preindustrial and the industrial periods separately, we observe in table 6 a significant decrease of 0.03 per decade in the association between father’s and son’s class during industrialization. Our fifth hypothesis is supported. Before the onset of industrialization, relative mobility slowly decreased. In other words, society became more closed. As appendix table A4 shows, the latter finding may be the result of changes in the percentage of cases with missing information on father’s class (model 2a) or may be driven by farmers’ immobility (model 2b). If we take either of these two variables into account 866 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility the trend in relative mobility in the preindustrial period becomes not significant. However, the trend toward more openness during industrialization remains significant and of the same size or stronger (20.03 in model 2a; 20.04 in model 2b). If we look at country-specific trends (fig. 3 and table 7), we can see that during industrialization Britain, Germany, and the Netherlands clearly show significantly decreasing association (i.e., an increase in openness) and no change or even increasing association in the decades before the onset of industrialization. Hungary and Sweden show the same pattern, but the coefficients are not significant (note that mobility tables—and not individuals—are the cases used in these analyses, and thus the sample size is small). In France the trend toward more openness is significant before industrialization but disappears thereafter. Finland does not show significant changes in association, and the direction of the changes contradicts what we expected. We performed a post hoc test of whether trends in relative mobility in France and Finland deviated significantly from the average trend in the other countries by adding interactions between (1) preindustrial time and the two countries and (2) industrial time and the two countries. The first two interactions have significant negative effects indicating that the decrease in relative mobility in the preindustrial period did not take place in France and Finland. The second two interactions have significant TABLE 7 Changes in Relative Mobility over Time Britain Finland France Germany Hungary Netherlands Sweden Model 1 Constant . . . . . . . . . . . Time . . . . . . . . . . . . . . Adj. R2. . . . . . . . . . . . . Model 2a Constant . . . . . . . . . . . Preindustrial time . . . . Industrial time . . . . . . Adj. R2. . . . . . . . . . . . . N................. 1.24** .87** 1.01** 2.04** 2.02 2.02** .65 .21 .63 .60** .00 .74 1.09** .00 .00 1.00** .01** .45 .60** .02** .42 1.06** .66** .86** .67** 2.01 2.03 2.03** .02* 2.04** .10 2.00 2.04* .63 .28 .70 .79 16 10 22 36 1.22** .09 2.10 .21 10 1.13** .03** 2.02** .89 21 .77** .02** 2.08 .44 19 NOTE.—Unstandardized coefficients of regression analysis of uniform difference parameters on time, using splines. The Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). a Model 2 contains splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. 867 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology positive effects, showing that in France relative mobility did not change at all and in Finland it even decreased. France and Finland appear to be outliers in this respect. Again we checked how robust our results are, given the differences between the data sets (app. table A5). First we investigate the influence of higher-order marriages. In the British data set and the Dutch Genlias data set higher-order marriages cannot be excluded. How does this affect the results? The data for Finland, France, Germany, and Hungary allow analyses with and without higher-order marriages. The four countries show no systematic differences between the two sets of results. We can conclude that support for the hypothesis of increasing relative mobility during industrialization in Britain and the Netherlands is not driven by the inclusion of higher-order marriages. We also compared results from the two Dutch data sets. The coefficients from the analyses do not differ much between the data sets, but they are not significant in the case of the HSN data, probably as a result of fewer tables and the smaller sample size within each table. The weighted Swedish data—containing fewer farmers’ sons—show slightly different coefficients, but the results are not substantially different from those using the unweighted data (the correlation between the two sets of mobility parameters is 0.98). All in all, the robustness checks do not give reason to doubt the conclusions from the main analyses. Overall, and in a majority of countries, relative mobility increased after the onset of industrialization but not before that time. We expected country differences in relative mobility to become smaller during industrialization (our sixth hypothesis). Just as in the case of total mobility, we calculated a set of coefficients of variation to test this hypothesis (see fig. 6). There is some support for the hypothesis of convergence among countries with respect to relative mobility patterns. After 1880 a decrease in variation is evident. However, the differences among countries do not become smaller than they were in the preindustrial period (between 1820 and 1860). The results would also be consistent with a temporary increase in country differences in relative mobility around the onset of industrialization. Finally, we decompose what is behind the trend toward increasing openness that we observed in most countries during industrialization. Does it occur because the likelihood to remain in the same class as one’s father decreases, or because for those who do leave the father’s class, the likelihood to move to specific destination classes becomes more equal? We estimate two uniform difference parameters, one for the association in the complete table, and one for the cells on the diagonal (see table 8). The trend toward less openness that we observed in Germany, the Netherlands, and Sweden before the onset of industrialization is mainly caused by an increase in the off-diagonal association. Both in the Netherlands and 868 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility FIG. 6.—Convergence and divergence of relative mobility in seven European countries, 1800–1914 (coefficients of variation). in Sweden immobility already decreased in the preindustrialization period, but this development was masked by an increasing off-diagonal association. Thus, not only in France, but also in these two countries, men were increasingly likely to end up in a different class from that of their father, even before industrialization started. However, their distribution over destination classes was increasingly unequal (after taking the different sizes of the classes into account). In general, the trend in mobility and that in immobility during industrialization were similar, but the changes in immobility were stronger and more significant than those in mobility. For example, in Britain the overall association between father’s and son’s class decreased with 0.04 per decade (P < 0.01), whereas immobility decreased with 0.09 (P < 0.05), and the offdiagonal association with 0.03 (P < 0.01). In Hungary the overall trend in relative mobility was not significant, but immobility significantly decreased during industrialization, and in the Netherlands the overall increase in openness seems completely caused by an increasing likelihood to leave the parental class. Taking these two observations together, we can conclude that changes in the likelihood for a son to remain in the parental class preceded changes in traversing over the whole social space and in some cases seem to have started already before industrialization, although in all countries but France industrialization gave these changes a boost. 869 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 1.22** .09 2.10 .21 O 1.06** 2.01 2.04** .63 I 1.26** .08 2.10 .22 10 M 1.21** 2.10* 2.17** .85 I 1.09** 1.15** .03 2.05 2.03* 2.09** .29 .48 16 HUNGARY M 1.13** .03** 2.02** .89 O .66** 2.03 .10 .28 O 1.17 .04** 2.00 .94 21 M NETHERLANDS .87** 2.01 2.00 .00 10 M FINLAND 1.12** 2.02** 2.10** .97 I 1.90** 2.01 2.44 .12 I .77 .02** 2.08 .44 O .86** 2.03** 2.00 .70 O .88** .01* 2.13 .22 19 M .91** 2.02** .00 .58 22 SWEDEN M FRANCE 1.33** 2.04* 2.29 .35 I .77** 2.07** 2.01 .52 I .67** .02* 2.04* .79 O .68** .02 2.00 .63 36 M GERMANY .74** .02 2.13 .68 I NOTE.—Unstandardized coefficients of regression analysis of uniform difference parameters on time, using splines. The Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). O 5 overall; M 5 mobility; I 5 immobility. a The models contain splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. Constant . . . . . . . . . . . Preindustrial time . . . . Industrial time . . . . . . Adj. R2 . . . . . . . . . . . . N................. Constant . . . . . . . . . . . Preindustrial time . . . . Industrial time . . . . . . Adj. R2 . . . . . . . . . . . . N................. O BRITAIN TABLE 8 Changes in Relative Mobility: Overall and Separately for Mobility and Immobility Trends in Male Intergenerational Class Mobility CONCLUSION We asked how total and relative mobility from father to son changed in European societies during industrialization. We derived hypotheses on total and relative mobility with regard to the amount of mobility within countries, differences between countries, and their convergence. We tested these hypotheses on a large and harmonized data set for seven European countries in the period 1800–1914. Taking all evidence together, qualified support is found for the industrialization hypotheses. Although country differences are not predicted well, the predicted increase of total and relative mobility at the onset of industrialization has been supported. In the following we discuss the results in more detail. An unexpected finding of our study is the similar development of total and relative mobility. Although the industrialization thesis predicts similar trends, a considerable body of previous research suggested differently. Total mobility was either expected to be rather unpredictable or to follow more or less directly from changes in class distribution—which of course cannot be true for relative mobility. Our study shows similar developments in both types of mobility over time and similar relative positions of countries with respect to both types of mobility. The industrialization thesis predicts that both total and relative mobility increased during the 19th century as a consequence of industrialization. In early industrializing countries (Britain), this should occur earlier than in countries that industrialized somewhat later (France, Germany, and the Netherlands) or much later (Finland, Hungary, and Sweden). Our study shows that both total and relative mobility increased in all countries together as well as in most countries separately, following the periodization predicted by the industrialization thesis. At the onset of industrialization total and relative mobility started to increase in five of the seven countries in our study (Finland and France being the odd ones out). Not only for total mobility, but also in the case of relative mobility, it was mainly the likelihood of sons to stay in exactly the same class as their father’s that decreased. These results suggest that the onset of industrialization may indeed have been a turning point. When we relate our findings to comparisons in the same tradition (Ganzeboom et al. 1989; Erikson and Goldthorpe 1992; Wong 1992; and Breen 2004) we can draw a tentative conclusion about changes in mobility on the very long run. Our conclusion is tentative, because we observed the period 1800–1914, and the other studies examined the period after the Second World War, while the interbellum remains unobserved. Furthermore the two sets of studies are not comparable in a strict sense. The results suggest that in Europe a trend toward more mobility took place that started during the Industrial Revolution and lasted until now. 871 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology Both in the 19th and in the second half of the 20th century, countries can be distinguished in which such a trend cannot be observed. In the 19th century, Finland and France deviated from the general pattern. In late industrializing Finland no changes in total and relative mobility are observed in our main analyses. We do, however, observe an increase in total mobility when we take into account that we cannot observe mobility of sons for which information on the father was not noted on the marriage act. Further research is necessary to discern whether these findings hold with more data and when investigating a longer period after the onset of industrialization. With around 4,000 cases, the Finnish data set is by far the smallest in our study. Most observations are from before the onset of industrialization, making it difficult to observe an effect. In France total and relative mobility increased after the beginning of the 19th century, but this increase slowed down during industrialization. In the literature it is sometimes suggested that it was not industrialization but the French Revolution that was the key historical event that opened up French society, although there is doubt as to whether it had an enduring effect (Sorokin 1959; Bourdieu et al. 2009). Our data do not permit us to test this hypothesis, since they do not extend as far back as the prerevolution period. However, our findings do show that, at the beginning of the 19th century, France experienced a trend toward increasing intergenerational mobility that lasted at least half a century. Although our study includes more countries than were examined in previous studies, data from even more countries are desirable to determine whether the observed trends in five out of the seven countries reflect the general pattern and France and Finland are deviations. A study covering a longer time period would also be advantageous to discern if these two countries permanently or only temporarily deviate from the other countries.14 The industrialization thesis fails to explain the sizable differences in total and relative mobility between countries. Sweden and Germany showed a high level of total and relative mobility. Whereas the results for Germany should be interpreted with care because of the poorer data quality, those for Sweden are surprising, since this is a late industrializing country. Rela- 14 Alternatively, one might compare our results with analyses of surveys on late industrializing countries. Whelan and Layte (2002), e.g., studied Ireland, Torche (2005), Chile, and Ishida (2003), Japan. According to these studies, in none of these countries did relative mobility change in the period of industrialization. Many of the problems connected with previous studies on 19th-century changes in mobility, however, also apply to these studies. The period studied by these authors is relatively short, the number of cases not too large, and differences in approach make the studies hard to compare among each other and with our study. These late industrializing countries are also more likely to differ from the countries in our study with respect to political regime, and they industrialize in a different context, experiencing competition from countries that industrialized much earlier. Adding more studies on 19th-century Europe thus still seems to be needed. 872 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility tive mobility was also relatively high in Finland and unexpectedly low in the Netherlands. Relative mobility in the Netherlands was comparable to that in Hungary, but Hungary had a much larger agricultural sector and started to industrialize 20 years later. Both in Hungary and in the Netherlands the low relative mobility seems to be mainly caused by the immobility of farmers. Why that is the case in two such different countries requires further investigation. Britain, the most industrialized country during the 19th century, took a middle position with respect to both total and relative mobility. For the 20th century Erikson and Goldthorpe (1992) too find that Britain (England and Scotland) took a middle position together with France. And also, in their study, total mobility rates were high in Sweden. Some of their findings deviate from ours, though. Germany does not show exceptionally high mobility in their study, which may be due to the poorer quality of our German data. Furthermore, we find a low level of total mobility in Hungary, whereas their study shows Hungary to be the most mobile of all their countries, possibly caused by the political intervention of the communist regime after the Second World War. The high level of total mobility in Sweden and the intermediate level of total mobility in Britain and France are, however, consistent over the 19th and a large part of the 20th century. Ganzeboom et al. (1989) present country-specific levels of relative mobility. In the 19th century relative mobility was relatively high in Germany, Sweden, and Finland, and low in Hungary and the Netherlands. In the 20th century this was still true for Finland and for the Netherlands. Hungary became relatively more open over time, in line with its higher total mobility rates in the 20th than in the 19th century. Germany, France, and Sweden form a middle group, and England was relatively open in the 20th century. If the level of industrialization is not clearly related to these country differences in total and relative mobility, how should these differences then be explained? One possibility is that the observed country differences are due to differences between the data sets, which are not completely identical. Some are more representative, some are larger, and some contain not only first but also later marriages. However, the main source in all countries are marriage records, and occupations have been coded into the same occupational classification and transferred to the same class schema. Furthermore we tested as best as we could the consequences of data quirks via several sensitivity analyses, and the results of the main analyses are robust. Together with the finding that 19th-century country differences are not so different from 20th-century country differences, this makes us believe that data issues are not the explanation for the failure of the data to comply with the theory. The previous comparisons of country differences suggest some explanations. Political regime change may play a role. This is suggested 873 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology both by the relatively high level of mobility in France in the first half of the 19th century, shortly after the French Revolution, and by the drastic changes in the relative position with respect to total and relative mobility in Hungary. Another explanation may be found in long and enduring country differences, such as in culture, religion, or political institutions. The high rates of mobility in Sweden and Finland (especially relative mobility in the case of Finland) and the low rates in the Netherlands seem to have lasted over two centuries. In Sweden and Finland, the Lutheran church may have played an important role. Large parts of the population in both Sweden and Finland were members of this church and were pressed, by the church and the state alike, to learn to read in order to understand the Bible (Johansson 1981). This may have opened up new occupational opportunities for lower-class people in these two countries. In the other countries in our study, no such reading campaign existed, at least not until much later, when elementary education became obligatory. This does not explain why, though, Dutch society was so closed, as this was also a literate country early on. While we have been able to demonstrate that total and relative mobility were on the rise at the same time as industrialization in European countries, we do not know whether industrialization was the cause. Alternatively, changes that took place at the same time and sometimes as a consequence of industrialization, such as educational expansion, the growth of communication, or an increase in travel, made it more likely for sons to obtain occupations different from their fathers. We think this study provides only qualified support for the industrialization thesis, because we do not yet know whether industrialization was the true cause of the changes or one of the concomitant processes. Furthermore, one country, France, shows a development over time against the predictions of the logic of industrialization theory, and country differences are not well predicted by the theory, even though trends in total and relative mobility among the first industrializing nations in the world are rather well predicted. All told, we believe the historical record available to us now suggests that in Europe a trend toward more social mobility took place that started during the Industrial Revolution and lasted until now. APPENDIX In figure A1, the solid lines represent the main analyses. Dashed lines represent robustness checks. 1. For Germany the solid line represents predicted scores (table 5, model 2) from all data sets combined. 2. For the Netherlands the main analyses pertain to the Genlias data set and the robustness check to the HSN data. 874 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility 3. For Sweden the robustness check uses data weighted to represent the original distribution of son’s class. 4. For Finland, France, and Hungary the robustness check includes higher-order marriages: figures are not shown, because the lines perfectly coincide. 875 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). FIG. A1.—Total mobility (proportions) in seven European countries, 1800–1914, main analyses and robustness checks. 876 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 877 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Type of Data Coverage Voland (2011) Mooser (1990) First marriages 6,902/4,651 First marriages 886/578 First marriages 15,415/8,076 HSN (2008) 1,036,193/509,344a Genlias (2009) Imhof (1998) First marriages 4,286/2,088 All marriages Knodel (1988) Kaelble and Federspiel (1990) Ditt (1990), Schüren (1990) Reference First marriages 7,190/3,651 First marriages 2,606/1,653 First marriages 3,851/1,791 N Original/ N Analyses b Second- and higher-order marriages cannot be excluded. The Genlias data come from the Regionaal Historisch Centrum Limburg (release June 7, 2006), Historisch Centrum Overijssel (release April 7, 2005), the Gelders Archief (release November 1, 2006), Zeeuws Archief (release June 28, 2004), Groninger Archieven (release March 16, 2007), and Drents Archief (release August 2010). c HSN: Historical Sample of the Netherlands. Data Set Civil Certificates Release 2008.01 a 1800–1914 Clan registers Complete population 1825–1909 Marriages (church) Random sample 1830–1914 Marriages (church) Sample stratified by social class (Bielefeld) 1800–1849 Clan registers Complete population of villages and cities 1800–1874 Marriages (church) Complete population of villages and cities Quernheim . . . . . 1800–1874 Marriages (church) Complete population of parish Netherlands: All marriages in 6 (out of 11) Genliasb . . . . . . . . 1810–1914 Marriages (civil) provinces Random sample of birth HSNc . . . . . . . . . . 1830–1914 Marriages (civil) certificates Netherlands Germany: Six Villages . . . . . Berlin . . . . . . . . . Bielefeld and Borghorst . . . . Hamburg and Villages . . . . . . Ost-Friesland . . . Period TABLE A1 Characteristics of the German and Dutch Data Sets 878 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). B 21.01** .02 .10 5,216 A 2.98** .02 .17 4,257 A 2.79** .06** .02 24,452 B 2.77 .06** .02 26,556 FRANCE 2.04 2.03* .06** 14,412 A 2.07 2.03** .06** 16,569 B GERMANY 2.93** 2.15** .23** 31,060 A 2.94** 2.15** .23** 34,036 B HUNGARY 2.25** 2.08** .03** 509,344 A 2.29** 2.09** .04* 8,076 C NETHERLANDS A 2.28** 2.03** .47** 23,384 D 2.19** 2.00 .28** 23,384 SWEDEN NOTE.—Coefficients (b) of logistic regression of intergenerational mobility on time, using splines. The main Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). A 5 original model, B 5 including higher-order marriages, C 5 alternative data set: HSN, D 5 weighed by original distribution of son’s class. The model contains splines for the effects of time in the preindustrial period and the industrial period, and a dummy variable distinguishing these two periods. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. Constant . . . . . . . . . . . Preindustrial time . . . . Industrial time . . . . . . N................. FINLAND TABLE A2 Robustness Checks of Changes in Total Mobility: In Preindustrial Years, and during Industrialization TABLE A3 Changes in Total Mobility over Time and with Taking Missing Information on Father’s Class into Account Britain Finland France Germany Hungary Netherlands Sweden Constant . . . . . . . . . 2.92** 2.78** 2.23 2.82** 22.15** 2.04 2.40* Preindustrial time . . . . . . . . . . . 2.05 .02 .04** .05 2.12** 2.08** 2.02** Industrial time . . . . .09** .26* .02 .04* .18** .02** .50** % father missing . . . .01 2.02* 2.01 .02** .03** 2.00** .00 Model v2 (df ). . . . . . 89.97 (3) 9.77 (3) 94.03 (3) 276.38 (8) 167.34 (3) 1247.29 (3) 31.30 (3) N . . . . . . . . . . . . . . . 9,836 4,257 24,452 14,412 31,060 509,344 23,384 NOTE.—Coefficients (b) of logistic regression of intergenerational mobility on time, using splines. The Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). The model contains splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. 879 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). TABLE A4 Changes in Relative Mobility over Time, with Missing Information on Father’s Class and without Farmer’s Immobility Constant . . . . . . . . . . . . . . . . . . . . . Time . . . . . . . . . . . . . . . . . . . . . . . . Preindustrial time . . . . . . . . . . . . . . Industrial time . . . . . . . . . . . . . . . . Britain . . . . . . . . . . . . . . . . . . . . . . . Finland . . . . . . . . . . . . . . . . . . . . . . France . . . . . . . . . . . . . . . . . . . . . . . Germany . . . . . . . . . . . . . . . . . . . . . Hungary . . . . . . . . . . . . . . . . . . . . . Netherlands . . . . . . . . . . . . . . . . . . Sweden . . . . . . . . . . . . . . . . . . . . . . % father missing . . . . . . . . . . . . . . . Adjusted R2 . . . . . . . . . . . . . . . . . . . N........................... Model 1a Model 2a Model 1b Model 2b 1.03** 2.01* 1.04** 1.00** 2.01** 1.01** .01 2.03** 2.18** .05 2.27** .26** .24** 2.10 2.00* .79 134 2.22** 2.03 2.34** .15** .14* 2.23** 2.00 .82 134 2.00 2.04** 2.08 .03 2.12* .16** .09* 2.34** 2.15** 2.02 2.17** .06 .04 2.42** .70 134 .73 134 NOTE.—Unstandardized coefficients of regression analysis of uniform difference parameters on country and time, using splines, all countries combined. The Dutch analyses pertain to the Genlias data set; for Germany all data sets are combined and the model is estimated with a dummy variable for each dataset (Six Villages data being the reference category). Missing data on father’s class are used in models 1a and 2a; models without farmers’ immobility are 1b and 2b. Model 2 contains splines for the effects of time in the preindustrial period and the industrial period. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). * P < .05. ** P < .01. 880 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 881 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). .66** 2.03 .10 .28 10 .65** 2.03* .13 .29 B .86** 2.03** 2.00 .70 A 22 B .86** 2.02** 2.00 .66 FRANCE .67** .02* 2.04* .79 A 36 .68** .01 2.03* .71 B GERMANY 1.22** .09 2.10 .21 A 10 1.18** .07 2.07 .28 B HUNGARY 1.13** .03** 2.02** .89 21 A 1.05** .02 2.03 .06 15 C NETHERLANDS .77 .02** 2.08 .44 A 19 D .75** .02* 2.03 .28 SWEDEN NOTE.—Regression analysis of uniform difference parameters on time, using splines. The main Dutch analyses pertain to the Genlias data set; for the German analyses all data sets are combined and the model is estimated with a dummy variable for each data set (Six Villages data being the reference category). The model contains splines for the effects of time in the preindustrial period and the industrial period, and a dummy variable distinguishing these two periods. All time effects are in decades (years/10). The industrial period is assumed to have started in 1850 (Britain), 1870 (France, Germany, Netherlands), or 1890 (Finland, Hungary, Sweden). A 5 original model, B 5 including higher-order marriages, C 5 alternative data set: HSN, D 5 weighed by original distribution of son’s class. * P < .05. ** P < .01. Constant . . . . . . . . . . Preindustrial time . . . Industrial time . . . . . Adjusted R2 . . . . . . . . N A FINLAND TABLE A5 Robustness Checks of Changes in Relative Mobility American Journal of Sociology REFERENCES Bairoch, Paul. 1982. “International Industrialization Levels from 1750 to 1980.” Journal of European Economic History 11:269–333. Berend, Ivan. 2013. An Economic History of Nineteenth-Century Europe: Diversity and Industrialization. Cambridge: Cambridge University Press. Blau, Peter M., and Otis Dudley Duncan. 1967. The American Occupational Structure. New York: Wiley. Blossfeld, Hans-Peter. 1987. “Entry into the Labor Market and Occupational Career in the Federal Republic: A Comparison with American Studies.” International Journal of Sociology 17:86–115. Bourdieu, Jérôme, Joseph P. Ferrie, and Lionel Kesztenbaum. 2009. “Vive la Différence? Intergenerational Mobility in France and the United States during the Nineteenth and Twentieth Centuries.” Journal of Interdisciplinary History 39:523– 57. Breen, Richard, ed. 2004. Social Mobility in Europe. Oxford: Oxford University Press. Breen, Richard, and Ruud Luijkx. 2004. “Social Mobility in Europe between 1970 and 2000.” Pp. 37–75 in Social Mobility in Europe, edited by Richard Breen. Oxford: Oxford University Press. Broadberry, Stephen, Rainer Fremdling, and Peter M. Solar. 2010. “Industry, 1700– 1870.” Pp. 164–86 in The Cambridge Economic History of Modern Europe. Vol. 1: 1700–1870, edited by Stephen Broadberry and Kevin H. O’Rourke. Cambridge: Cambridge University Press. Broadberry, Stephen, and Kevin H. O’Rourke, eds. 2010a. The Cambridge Economic History of Modern Europe. Vol. 1: 1700–1870. Cambridge: Cambridge University Press. ———. eds. 2010b. The Cambridge Economic History of Modern Europe. Vol. 2: 1870 to the Present. Cambridge: Cambridge University Press. CEDAR (Centre for Demographic and Ageing Research) 2015. Popum Database. http:// www.cedar.umu.se/english/ddb/databases/popum/ (accessed June 24, 2015). Crafts, Nick F. R., and C. K. Harley. 1992. “Output Growth and the British Industrial Revolution: A Restatement of the Crafts-Harley View.” Economic History Review 45:703–30. Davis, Kingsley. 1955. “Social and Demographic Aspects of Economic Development in India.” Pp. 263–315 in Economic Growth: Brazil, India, Japan, edited by S. Kuznets, W. E. Moore, and J. J. Spengler. Durham, N.C.: Duke University Press. Ditt, Karl. 1990. Family and Social Placing in the Bielefeld Underclasses in the 19th Century. GESIS Data Archive, Cologne. ZA8021 Data file ver. 1.0.0. doi:10.4232/1.8021. Dupâquier, Jacques. 2004. “L’enquête des 3,000 familles.” Annales de Démographie Historique 1:7–18. Erikson, Robert. 1983. “Changes in Social Mobility in Industrial Nations: The Case of Sweden.” Research in Social Stratification and Mobility 2:165–95. Erikson, Robert, and John H. Goldthorpe. 1987. “Commonality and Variation in Social Fluidity in Industrial Nations. Part II: The Model of Core Social Fluidity Applied.” European Sociological Review 3:145–66. ———. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Oxford University Press. Erikson, Robert, John H. Goldthorpe, and Lucienne Portocarero. 1979. “Intergenerational Class Mobility in Three Western European Societies: England, France and Sweden.” British Journal of Sociology 30:415–51. ———. 1982. “Social Fluidity in Industrial Nations: England, France and Sweden.” British Journal of Sociology 33:1–34. ———. 1983. “Intergenerational Class Mobility and the Convergence Thesis: England, France and Sweden.” British Journal of Sociology 34:303–43. 882 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility Featherman, David L., F. Lancaster Jones, and Robert M. Hauser. 1975. “Assumptions of Social Mobility Research in the U.S.: The Case of Occupational Status.” Social Science Research 4:329–360. Form, William. 1979. “Comparative Industrial Sociology and the Convergence Hypothesis.” Annual Review of Sociology 5:1–25. Fukumoto, Ivan K., and David B. Grusky. 1993. “Social Mobility and Class Structure in Early Industrial France.” Pp. 40–67 in Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940, edited by Andrew Miles and David Vincent. Manchester: Manchester University Press. Ganzeboom, H., R. Luijkx, J. Dessens, P. de Graaf, N. D. de Graaf, W. Jansen, and W. Ultee. 1987. “Intergenerationele klassenmobiliteit in Nederland tussen 1970 en 1985.” Mens en Maatschappij 62:17–43. Ganzeboom, Harry B. G., Ruud Luijkx, and Donald J. Treiman. 1989. “Intergenerational Class Mobility in Comparative Perspective.” Research in Social Stratification and Mobility 8:3–84. Ganzeboom, Harry B. G., Donald J. Treiman, and Wout C. Ultee. 1991. “Comparative Intergenerational Stratification Research: Three Generations and Beyond.” Annual Review of Sociology 17:277–302. Genlias. 2009. Genlia— Zoek uw stamboom uit. Retrieved October 11, 2009. http://www .genlias.nl/ [website since replaced by https://www.wiewaswie.nl]. Goldthorpe, John H., and Lucienne Portocarero. 1981. “La mobilité social en France, 1953–1970: Nouvel examen.” Revue Française de Sociologie 22:151–66. Grafe, Regina, Larry Neal, and Richard W. Unger. 2010. “The Services Sector.” Pp. 187– 216 in The Cambridge Economic History of Modern Europe. Vol. 1: 1700–1870. Edited by Stephen Broadberry and Kevin H. O’Rourke. Cambridge: Cambridge University Press. Häkkinen, A. 2006. Ten Generations—Three Centuries: A Finnish History as Family Stories. http://blogs.helsinki.fi/hakkinen/tutkimus/ten-generations-three-centuries-a -finnish-history-as-family-stories/ Accessed October 2016. Hajnal, John. 1953. “The Marriage Boom.” Population Index 19:80–101. ———. 1965. “European Marriage Patterns in Perspective.” Pp. 101–43 in Population in History: Essays in Historical Demography, edited by David V. Glass and D. E. C. Eversley. London: Edward Arnold Publishers. Hout, Michael. 1988. “More Universalism, Less Structural Mobility: The American Occupational Structure in the 1980s.” American Journal of Sociology 93:1358– 1400. Hout, Michael, and Avery M. Guest. 2013. “Intergenerational Occupational Mobility in Great Britain and the United States since 1850: Comment.” American Economic Review 103:2021–40. HSN (Historical Sample of Netherlands). 2008. Data Set Civil Certificates Release 2008.01 http://www.iisg.nl/hsn/. ———. 2010. Data Set Life Courses Release 2010.01. http://www.iisg.nl/hsn/. Imhof, Arthur E. 1998. Life-Expectancy in Germany, 1700 to 1890. GESIS Data Archive, Cologne. ZA8066 Data file ver. 1.0.0. doi:10.4232/1.8066. Inkeles, Alex. 1960. “Industrial Man: The Relation of Status to Experience, Perception, and Value.” American Journal of Sociology 66:1–31. Ishida, Hiroshi. 2003. “Industrialization, Class Structure, and Social Mobility in Postwar Japan.” British Journal of Sociology 52:579–604. Ishida, Hiroshi, John H. Goldthorpe, and Robert Erikson. 1991. “Intergenerational Class Mobility in Postwar Japan.” American Journal of Sociology 96:954–92. Johansson, Egil. 1981. “The History of Literacy in Sweden.” Pp. 151–83 in Literacy and Social Development in the West: A Reader, edited by Harvey J. Graff. Cambridge: Cambridge University Press. 883 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). American Journal of Sociology Jonsson, Jan O., and Colin Mills. 1993. “Social Mobility in the 1970s and 1980s: A Study of Men and Women in England and Sweden.” European Sociological Review 9:229–47. Kaelble, Hartmut. 1981. Historical Research on Social Mobility: Western Europe and the USA in the Nineteenth and Twentieth Centuries. London: Croom Helm. ———. 1985. Social Mobility in the 19th and 20th Centuries: Europe and America in Comparative Perspective. Leamington Spa: Berg. Kaelble, Hartmut, and Ruth Federspiel, eds. 1990. Historische Statistik von Deutschland. Band X. Soziale Mobilität in Berlin. 1825–1957. St. Katharinen: Scripta Mercaturae Verlag. Kerr, Clark, John T. Dunlop, Frederick H. Harbison, and Charles A. Myers. (1960) 1973. Industrialism and Industrial Man. Harmondsworth: Penguin. Knodel, John E. 1988. Demographic Behavior in the Past: A Study of Fourteen German Village Populations in the Eighteenth and Nineteenth Centuries. Cambridge: Cambridge University Press. Lambert, Paul S., Kenneth Prandy, and Wendy Bottero. 2007. “By Slow Degrees: Two Centuries of Social Reproduction and Mobility in Britain.” Sociological Research Online 12 (1). Leroy, P., D. Nicolas, and J. P. Pélissier. 1998. “Mobilités et aires matrimoniales.” Pp. 61– 75 in Les chemins de la recherche: Le choix du conjoint, edited by G. Brunet, A. FauveChamoux, and M. Oris. Lyon: Programme Rhône-Alpes de Recherches en Sciences Humaines. Lippényi, Zoltán, Ineke Maas, and Marco H. D. van Leeuwen. 2013b. “Intergenerational Class Mobility in Hungary between 1865 and 1950: Testing Models of Change in Social Openness.” Research in Social Stratification and Mobility 33:40–55. ———. 2013a. “Creating the Hungarian Historical Social Mobility File: Historical Social Structure and Mobility beyond the Leitha.” Istoricheskaya Informatika 2:3–23. Lipset Seymour M., and Hans L. Zetterberg. 1959. “Social Mobility in Industrial Societies.” Pp. 11–75 in Social Mobility in Industrial Society, edited by R. Bendix and S. M. Lipset. Berkeley: University of California Press. Long, Jason, and Joseph Ferrie. 2013. “Intergenerational Occupational Mobility in Great Britain and the United States since 1850.” American Economic Review 103: 1109–37. Maas, Ineke, and Marco H. D. van Leeuwen. 2002. “Industrialization and Intergenerational Mobility in Sweden.” Acta Sociologica 45:179–94. Maas, Ineke, Marco H. D. van Leeuwen, Jean-Pierre Pélissier, and Danièle Rébaudo. 2011. “Economic Development and Parental Status Homogamy: A Study of 19th Century France.” History of the Family 16:371–386. Miles, Andrew. 1993. “How Open was Nineteenth-Century British Society? Social Mobility and Equality of Opportunity, 1839–1914.” Pp. 18–39 in Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940, edited by Andrew Miles and David Vincent. Manchester: Manchester University Press. ———. 1999. Social Mobility in Nineteenth- and Early Twentieth-Century England. Basingstoke: Macmillan. Miles, Andrew, and David Vincent, eds. 1993. Building European Society: Occupational Change and Social Mobility in Europe, 1840–1940. Manchester: Manchester University Press. Mooser, Josef. 1990. Familie und soziale Plazierung in der ländlichen Gesellschaft am Beispiel des Kirchspiels Quernheim im 19. Jahrhundert. GESIS Data Archive, Cologne. ZA8031 Data file ver. 1.0.0. doi:10.4232/1.8031. Persson, Karl Gunnar. 2010. An Economic History of Europe: Knowledge, Institutions and Growth, 600 to the Present. Cambridge: Cambridge University Press. Poppel, Frans van, Jurjen de Jong, and Aart C. Liefbroer. 1989. “The Effects of Paternal Mortality on Sons’ Social Mobility: A Nineteenth-Century Example.” Historical Methods: A Journal of Quantitative and Interdisciplinary History 31:101–12. 884 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Trends in Male Intergenerational Class Mobility Schüren, Reinhard. 1989. Soziale Mobilität: Muster, Veränderungen und Bedingungen im 19. und 20. Jahrhundert. St. Katharinen: Scripta Mercaturae Verlag. ———. 1990. Family and Social Placing in a Parish Molded by Agriculture, Cottage Trade and Industrialization with the Example of the Parish Borghorst in the 19th Century. GESIS Data Archive, Cologne. ZA8037 Data file ver. 1.0.0. doi:10.4232/1.8037. Schulz, Wiebke. 2013. “Careers of Men and Women in the 19th and 20th Centuries.” PhD dissertation. University of Utrecht, Interuniversity Centre for Social Science Theory and Methodology. Simkus, Albert. 1984. “Structural Transformation and Social Mobility: Hungary 1938– 1973.” American Sociological Review 49:291–307. Sorokin, Pitirim A. (1929) 1959. Social and Cultural Mobility. New York: Free Press. Statistique Générale de la France. 1926. Annuaire Statistique, vol. 42. Paris: Imprimerie Nationale. Thêlot, Claude. Tel père, tel fils ? Position sociale et origine familiale. Paris: Dunod. Torche, Florencia. 2005. “Unequal but Fluid: Social Mobility in Chile in Comparative Perspective.” American Sociological Review 70:422–50. Treiman, Donald J. 1970. “Industrialization and Social Stratification.” Pp. 207–34 in Social Stratification: Research and Theory for the 1970s, edited by Edward O. Laumann. Indianapolis: Bobbs-Merrill. Van Leeuwen, Marco H. D., and Ineke Maas. 2010. “Historical Studies of Social Mobility and Stratification.” Annual Review of Sociology 36:429–51. ———. 2011. HISCLASS: A Historical International Social Class Scheme. Leuven: Leuven University Press. Van Leeuwen, Marco H. D., Ineke Maas, and Andrew Miles. 2002. HISCO: Historical International Standard Classification of Occupations. Leuven: Leuven University Press. ———. 2004. “Creating a Historical International Standard Classification of Occupations: An Exercise in Multinational Interdisciplinary Cooperation.” Historical Methods 37:186–97. Van Leeuwen, Marco H. D., Ineke Maas, Danièle Rébaudo, and Jean-Pierre Pélissier. 2016. “Social Mobility in France, 1720–1986: Effects of Wars, Revolution and Economic Change.” Journal of Social History 49:585–616. Vermunt, Jeroen K. 1997. LEM: A General Program for the Analysis of Categorical Data. Tilburg University, Department of Methodology and Statistics. Voland, Eckart. 2011. “Die Familien der Krummhörn (Ostfriesland) des 18. und 19. Jahrhunderts, Datenbankbeschreibung.” Handbook. Version of May 31. Whelan, Christopher T., and Richard Layte. 2002. “Late Industrialization and the Increased Merit Selection Hypothesis: Ireland as a Test Case.” European Sociological Review 18:35–50. Wong, Raymond Sin-Kwok. 1992. “Vertical and Nonvertical Effects in Class Mobility: Cross-National Variations.” American Sociological Review 57:396–410. Wong, Raymond Sin-Kwok, and Robert M. Hauser. 1992. “Trends in Occupational Mobility in Hungary under Socialism.” Social Science Research 21:419–44. Xie, Yu. 1992. “The Log-Multiplicative Layer Effect Model for Comparing Mobility Tables.” American Sociological Review 57:380–95. Xie, Yu, and Alexandra Killewald. 2013. “Intergenerational Occupational Mobility in Great Britain and the United States since 1850: Comment.” American Economic Review 103:2003–20. Zijdeman, Richard L. 2009. “Like My Father before Me: Intergenerational Occupational Status Transfer during Industrialization (Zeeland, 1811–1915).” Continuity and Change 24:455–486. 885 This content downloaded from 131.211.104.035 on January 17, 2017 03:18:58 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
© Copyright 2025 Paperzz