Wealthier and healthier? The effect of socioeconomic status on - - PDF document

1

Wealthier and healthier?

The effect of socioeconomic status on later-life mortality in two 19th-century cohorts

Rick Mourits, Radboud University Nijmegen, the Netherlands. Email: r.mourits@let.ru.nl Ken R Smith, University of Utah, USA. Email: ken.smith@fcs.utah.edu Angelique Janssens, Radboud University Nijmegen, the Netherlands. Email: a.janssens@let.ru.nl The authors would like to thank Evan Roberts and Kees Mandemakers for their vital support in translating the Occ1950 and HISCO codes to the different class coding schemes. Wordcount: 6,537 (8,000)  591  1567  1901  1321  1156 Keywords (0/6) 19th century, socioeconomic status, later-life morality, inequality

2 Abstract 150-250 Studies on contemporary populations have shown that socioeconomic status is strongly related with

• ne’s life expectancy. Yet, little is known about the relationship between socioeconomic status and

individual survival in the past, as research has focused on when social inequalities started appeared. Although SES is thought to be one of the most important predictors of survival for 20th-century cohorts, the effect seems absent for those who lived and died before WWII. Recently, Temby and Smith (2014) found a link between longevity and socioeconomic status for 19th-century cohorts in Utah after they controlled for similarities between siblings. In this paper, we study the robustness of Smith affect the link between socioeconomic status and mortality after age 50 in the Utahn 1860- 1890 cohort and the 1812-1862 cohort from the Dutch province Zeeland.

3

• 1. Introduction

Stratification of health by social standing and income is one of the axioms in social science (Elo, 2009). Higher socioeconomic status significantly increases an individual’s survival chances in later- life, culminating in longer and healthier lifespans for those at the top of the social pyramid (Antonovksy, 1967; Bobak, Pikhart, Rose, Hertzman, & Marmot, 2000; Chen, Yang, & Liu, 2010; Huisman, Kunst, & Mackenbach, 2003; Kunst et al., 2004; Mackenbach et al., 2008; Zimmer, Hanson, & Smith, 2016a). Individuals from the lowest income and educational groups are 1.2 to 1.8 times as likely to report that they live in poor health than the highest social groups (Mackenbach et al., 2008; Zimmer et al., 2016a). Moreover, the death rate of the lowest social class is 1.5 to 2 times as high as the death rate of the highest social class in Western Europe (Mackenbach et al., 2008), culminating in a lower life expectancy of two to three years for men and about one year for women (Kunst et al., 2004). Yet, these patterns might be unique for contemporary societies, as class-based disparencies in later-life expectancy seem to have been different for individuals born before the 20th century (Bengtsson & van Poppel, 2011; Edvinsson & Broström, 2012). Whether the wealthier have always been healthier is a long-standing discussion. A wide collection

• f studies on historical socioeconomic differentials in infant mortality, child mortality, and adult

mortality provides little evidence of a socioeconomic graduent in mortality (Bengtsson & van Poppel, 2011). Recently, multiple studies shifted their focus to class differences in later-life mortality. Most of these papers showed that mortality risks in later life were elevated for the elite and lower for farmers in 19th-century cohorts (Edvinsson & Broström, 2012; Ferrie, 2003; Gagnon, Tremblay, Vézina, & Seabrook, 2011; Schenk & van Poppel, 2011; Smith, Mineau, Garibotti, & Kerber, 2009; Temby & Smith, 2014). The relationship between socioeconomic status and later-life survival is still less clear when examined across time and place. In Utah, there was a positive effect of socioeconomic status on survival after age 50 and the chance of belonging to the top 5% of survivors (Smith et al., 2009; Temby & Smith, 2014). Similarly, survival in later life increased linearly by

• ccupation in Québec and by income in Sart, Belgium (Alter, Neven, & Oris, 2004; Gagnon et al.,

2011). Yet, other papers on Belgium, the Netherlands, and Sweden have not been able to establish a connection between social position and later-life mortality (Bengtsson & Dribe, 2011; Donrovich, Puschmann, & Matthijs, 2014; Edvinsson & Broström, 2012; Schenk & van Poppel, 2011). Hitherto, different papers have applied different measures of social class, rendering it hard to determine whether diverging results were caused by methodological decisions or actual differences. In this paper we enquire whether the different findings were caused by spatial and temporal differences or methodological decisions. Eight different social class schemata are used to test the positive relationship between socioeconomic status and later-life survival that Smith et al. (2009) and Temby and Smith (2014) found for 19th-century cohorts. The same results are also produced for 19th- century cohorts from the Dutch coastal province of Zeeland. Combined, the datasets offer a rare insight in the 19th-century population dynamics of two radically different populations: a fledgling industrializing society on the American frontier and an established commercial-agricultural society in the old world. We ask ourselves two questions: (1) What is the relationship between socioeconomic status and later-life mortality in Utah and Zeeland? and (2) To what extent are results affected by methodological choices, i.e. the operationalization of SES? 2.1 Socioeconomic status and later-life mortality The lives of individuals born in the 19th century were rather different from the lives that we live

• today. The 19th century was an age of transition: the western world industrialized, developed the

4 basics of modern and preventive medicine, and revolutionized the infrastructure for transport and

• communication. These developments caused fundamental changes in the labor market (Van

Leeuwen & Maas, 2007), altered the disease environment (Omran, 1971), triggered a demographic transition (Davis, 1945; Dyson, 2010; Notestein, 1945), and allowed people to live longer (Kannisto, 2001; Mourits, 2017; Robine & Paccaud, 2005; Wilmoth, 2000). Yet, how these changes affected the relationship between socioeconomic status and mortality in later life is still not fully understood. Class-based inequalities in later-life mortality indicate that increased social standing comes with the ability to postpone death. In contemporary societies, higher social strata are able to transform material and immaterial socioeconomic resources into health. Causal pathways between socioeconomic status and later-life survival have been established for health-related behavior, such as healthy diets and lower tobacco consumption, although there are indications that the relationship is more complex and multifaceted (Christensen & Vaupel, 1996; Elo, 2009; Schrijvers, Stronks, Van de Mheen, & Mackenbach, 1999). Whether class-based differences in health also existed in the past is subject of a longstanding debate. Mortality in the 19th century was largely determined by malnourishment and infectious diseases rather than by obesity and degenerative diseases. Medical treatments and preventive healthcare against infectious diseases were generally not very effective, so that decreased exposure to pathogens and a healthy immune system were the main lines of defense against disease (McKeown, 1976; Rotberg & Rabb, 1985). Therefore, a causal link between socioeconomic status and later-life survival could not have been the same for contemporary and 19th-century populations. Infectious diseases dramatically affected all in the 19th century. In later-life, the death rate due to degenerative diseases such as cancer, cardiovascular, and chronic diseases was high, but tuberculosis, diarrhea, respiratory, parasitic and other infectious diseases were the main causes of death (Preston, 1976). All social classes were subjected to the same pathogens, as poor and rich lived right beside one another and access to clean drinking water was limited. Because avoiding sources of infection was practically impossible, Edvinsson and Broström poised the question whether “much of what was required to prevent ill health was in reach for less affluent groups as well” (2012, p. 653). Cynically, one could argue that wealth could only buy products that were detrimental to human health, e.g. coffee, tobacco and sugar (Brooshooft, 1897; Razzell & Spence, 2006). However, income also correspondend with daily nourishment, marginal improvements in housing quality, and how much one could spend on petrol, cokes, and wood to heat the room (Brooshooft, 1897). Daily nutrition is often thought to be the most important determinant to prevent and survive infection (McKeown, 1976; Rotberg & Rabb, 1985). Research on human stature has shown that differences in wealth caused visible differences in height between social classes. Elite, middle class, and rich farmer childeren were on average taller than their fellow countrymen, whereas unskilled workers lacked in stature (see e.g. Beekink & Kok, 2017; Mazzoni, Breschi, Manfredini, Pozzi, & Ruiu, 2017; Ramon-Muñoz & Ramon-Muñoz, 2017). Nevertheless, the relationship between nourishment and later-life survival is not as straightforward as one would assume. By stunting growth, the human body adapted to the available number of calories to prevent energy shortages and malfunction of bodily functions (Floud, Fogel, Harris, & Hong, 2011). As a result, the effect of calorie intake on one’s immune system was restricted. Moreover, some diseases – e.g. smallpox, malaria, typhoid, tetanus, and malaria – were so infectious that nutrition could not prevent infection. Hence, a chronic calorie shortage was not automatically a pathway to an untimely demise. Yet, nourishment is more than the sum of calories consumed. Insecurity of food and vitamin deficiencies were harder for the body to adapt to and probably affected the immune system

5

• negatively. Class-based differences in vitamin intake were certainly present. The poorest laborers

could barely make ends meet and survived on the bare essentials: bread, potatoes, rice, and some fat and milk. The average laborer also consumed bread, potatoes, fat, and milk, but could further afford beans, vegetables, and butter, while the most skilled laborers were able to buy meat or bacon twice a week (Brooshooft, 1897). Furthermore, higher social classes were more secure of food and income, as their work was less dependent on the seasons. This put a strain on the poorest, whose diets were already meager to start with, whereas higher classes and farmers fared better due to their secured access to food. As such, class-based inequalities in chronic vitamin and temporal calorie deficiencies might have been too large for the human body to overcome and affect physical fitness and the immune system. Therefore, we hypothesize that (1a) Individual socioeconomic status decreased mortality risks in later life, and (1b) Mortality risks in later life were lower for farmers. Yet, food was not scarce for all and some might have suffered from overabundance. The upper echelons in society had an estimated diet of over 3,000 calories per day (Floud et al., 2011, p. 53). Combined with a lack of physical activity, and excessive consumption of alcohol and tobacco, the behavior of the wealthiest was a mix of unhealthy practices (Razzell & Spence, 2006). The outcomes

• f these practices were well-known in the 1950s, when the upper classes were more likely to die of

ischaemic heart disease, lung cancer, diabetis mellitus, hypertensive disease, and cerebrovascular

• disease. These man-made diseases were seen as “diseases of affluence”, and somewhat reduced the

already existant, inversed effect of socioeconomic status on morality risks (Kunst, Looman, & Mackenbach, 1990). The positive correlation between man-made diseases and socioeconomic status inversed in the 1960s, when consumption of tobacco, alcohol, and sugar increased in the lower classes (Marmot, Adelstein, Robinson, & Rose, 1978). Until then, consumption of these products was a privilige of the upper classes and might have shortened their lives considerably. Therefore, we test whether (2) Mortality risks in later life were elevated for individuals from the highest social classes. 2.2 Earlier results and measurement Findings on the relationship between socioeconomic status and mortality risks in later life have been

• mixed. Smith et al. (2009) and Temby and Smith (2014) found a linear relation between

socioeconomic status and later-life mortality risks in Utah for men born between 1840 and 1909. Similarly, Gagnon et al. (2011) report a linear decrease in mortality risks between unskilled, semiskilled, and skilled laborers in Québec males born before 1875, and Alter, Neven & Oris (2004) show that wealth decreases the mortality risk after age 40 for men born in the Belgian town of Sart between 1815-1828. No class-based inequalities in mortality were found for the Antwerp region and southern Sweden (Bengtsson & Dribe, 2011; Donrovich et al., 2014). Other studies found different survival patterns. Farmers had better survival rates in Illinois, Mississippi, the Netherlands, Québec, and Utah (Edvinsson & Broström, 2012; Ferrie, 2003; Gagnon et al., 2009; Schenk & van Poppel, 2011; Temby & Smith, 2014), whereas mortality risks were elevated for the upper class in Mississippi, northern Sweden, the Netherlands, and Québec at the turn of the 19th century (Edvinsson & Broström, 2012; Ferrie, 2003; Gagnon et al., 2011; Schenk & van Poppel, 2011). Whether studies are able to find class-based differences in later-life survival seems to correspond with two methodological choices. First, whether men and women are studied separately. Bengtsson and Dribe (2011) and Donrovich, Puschmann & Matthijs (2014) studied men and women simultaneously and failed to find significant results. The number of occupations in which women worked was severely limited and female employment was often not registered (Janssens, 2014; Walhout & van Poppel, 2003). Moreover, effects of a husband’s socioeconomic status on female

6 later-life survival seem to be mediated by gender norms. Farmer’s wives had for example no extended survival, whereas women who were married to a craftsman had a better later-life expectancy (Alter, Dribe, & Van Poppel, 2007; Edvinsson & Broström, 2012; Schenk & van Poppel, 2011; Zimmer, Hanson, & Smith, 2016b). Thus, male and female later-life expectancy are two fields that should be studied separately. Second is the measurement of socioeconomic status. Different regions are studied with different socioeconomic schemata. The Antwerp region is studied with a collapsed HISCLASS scale (Donrovich et al., 2014), Illinois and Mississippi with household income and real estate assets (Ferrie, 2003), Sart with parental income (Alter et al., 2004), Sundsvall with a local class scheme (Edvinsson & Broström, 2012), Québec with SOCPO (Gagnon et al., 2011), Scania with a SOCPO scale that does not distinguish farmers (Bengtsson & Dribe, 2011), and Utah with the Nam-Powers scale (Smith et al., 2009; Temby & Smith, 2014; Zimmer et al., 2016b). The variation in social class schemata makes it hard to compare results, as similar occupations are scored differently between regions. This is especially troublesome as socioeconomic schemata were never created to stratify health outcomes. Methodologically, social class schemes treat the ability to transform one’s socioeconomic status into a longer and healthier life as a by-product, rather than an indicator, of social stratification. As such, the relationship between socioeconomic status and later-life expectancy might depend quite strongly on how socioeconomic status is measured. Therefore, we hypothesize that: (3) The relationship between individual socioeconomic status and later-life survival is dependent on the social class scheme.

• 3. Data and methods

This study uses data on 19th-century birth cohorts from Utah and the Dutch province of Zeeland. Utah and Zeeland represent important and useful populations for this study for multiple reasons. First, they reflect an industrializing society on the American fronteer and a European commerical- agricultural society in terms of social standing and adult mortality risk, thus allowing for a robust comparison between these broad populations. Second, each has been studied but relied on measures of socioeconomic status that developed separately on the two continents. Third, the two databases offer a rare insight in 19th-century population dynamics and have been exposed to rigorous quality control processes so as to enhance the quality of the comparisons and minimize artifactual differences attributable to varying levels of data quality. Data on Utahns reflect the experiences of living in a frontier region of the United States (Utah became a state in 1896), situated in an arid climate generally more than 1,300 metres (4,000 feet) above sea level. Prior to 1847, the land was inhabited by several American Indian tribes. Most early immigrants to the Utah territory were members of the recently established Church of Jesus Christ of Latter-day Saints (Mormons), although many non-Mormons arrived in the state as well. From their headquarters in Salt Lake City, the Mormon church sought to establish their influence over the area by coordinating settlements in unpopulated locales. Other streams of immigrants involving mining and later railroads, with the opening of the transcontinental railroad in 1869 as memorialized by the Golden Spike that linked the nation. The population of Utah became more diversified until in the 1920s immigration receded due to the economic downturn (Bean, Mineau, & Anderton, 1990). Utah will be compared to the Dutch province of Zeeland. Zeeland is a coastal province consisting

• f multiple islands, situated in the southwest of the Netherlands. The island archipelago had been
• ne of the richer regions of the Netherlands and home to the WIC in early modern times, but

suffered economic decline in the 18th century, so that the overseas trading and shipping industry disappeared at the dawn of the 19th century. Although the most populated islands were connected to

7 the central railway network by 1873, most of the province was poorly connected to the rest of the

• Netherlands. The province never industrialized during the 19th century, as the economic focus had

shifted to cash crops. Nevertheless, the growth of the cash crop sector was short-lived, as the cash crop sector plummeted at the turn of the 20th century (Priester, 1998; Van Zanden & Van Riel, 2000). As a result, Zeeland was a region characterized by outmigration and little population growth. The cohorts that grew up in Utah and Zeeland both lived in a rural society at the dawn of industrialization and grew old before the 20th century. However, the two regions have noteworthy

• differences. As a coastal region with few sources of clean drinking water, clayey soils and a high

humidity, Zeeland was characterized by a high mortality regime where between 30% and 40% of all newborns died before age five. In Utah, the comparable percentage was just over 10%. Nevertheless, average survival after age 50 was lower for men in Utah than in Zeeland, whereas women lived about as long. In 1860 the mean age at death for 50-year olds was at age 50 was 73.2 for men and 74.5 for women contrary to 74.3 and 74.1 in Zeeland. However, the 10, 5 and 1% top survivors of the 1860 cohort had a much higher average age-at-death (men: 90.2, 93.1, 98.1 / women: 90.1, 93.1, 97.8) than in Zeeland, (men: 89.5, 91.8, 95.8 / women: 89.6, 91.8, 95.6). The agricultural tradition and labor market on the Utah frontier differed from what was common in Zeeland. Between 1860 and 1940, Utah transformed from an agricultural to an industrialized

• society. In 1880 one in three Utahn males was a farmer, decreasing to one in seven by 1940. Farms

were generally family businesses and farm laborers were generally family members. As a result, only a small group of men worked as a farm worker or general worker over their entire lifespan. In Zeeland, on the other hand, the labor market was geared towards commercial agriculture and changed little over time. About 50% of all working men was a farm laborer or laborer and worked for farmers who comprised 10-15% of the population. The agricultural produce was sold and transported by shippers and cattle dealers, who were a sizable contingent of the property owners and managers in Zeeland. As a result, semi-skilled, service, and white collar labor was much less common in Zeeland. Figure 1: Labor market Utah& Zeeland Sources Data on Utah is drawn from the Utah Population Database (UPDB), which includes demographic and genealogical data for individuals born in Utah or on the Mormon Pioneer Trail and their descendants.The original portion of the UPDB is based on family group sheets kept by the Church of

8 Jesus Christ of Latter-Day Saints. The Church reconstructed the demographic history of Utah by compiling three-generation family sheets (parents of ego, ego/spouse, offspring) for individuals who had experienced at least one vital event (birth, marriage of death) in Utah or while immigrating along the Mormon trail. With this information, obtained from the Utah Genealogical Society, basic demographic information of 1.6 million individuals – containing data on Mormons and non-Mormons as well as in- and out-migrants – was reconstructed. Information on occupational status was not kept in these records, but additional information on socioeconomic status has been linked to the UPDB from death certificates (1904-present) and the 1880, 1900, 1910, 1920, 1930 and 1940 Utah Census. Data on Zeeland were retrieved from the Dutch civil registry of Zeeland with the Linking System for Historical Family Reconstruction (Kees Mandemakers & Laan, 2017). The Dutch civil registry is one

• f the oldest in the world and of high quality. Municipalities were required to register all vital events

in a very precise manner. As a result, all vital events recorded over the last 200 years can be retraced. By linking the egos on birth, marriage, and death certificates, individual lifespans were reconstructed, whereas families were reconstructed by linking parents on birth and marriage certificates to egos. The resulting dataset contains complete lifespan and family relations for individuals who were born in Zeeland between 1812 and 1862, married in Zeeland between 1812 and 1937, and died in Zeeland between 1812 and 1962. Information on occupational status is available on an individual’s death certificate, as well as on the marriage and death certificates of his/her children. Socioeconomic status For our analyses, we included male cohort members who died after age 50 and for whom information on the occupational career and family of origin was available. Women were excluded from the analyses as female occupational employment was registered only minimally during the 19th century (Janssens, 2014; Walhout & van Poppel, 2003). For Utah, the occupational career of 44,751

• ut of 88,895 male members the 1860-1890 cohort who lived to be at least 50 years old were

followed through linked census records from 1880, 1900, 1910, 1920, 1930, and 1940. Moreover, information on an individual’s most prominent, i.e. longest held, occupation was available from 1904

• nwards on death certificates. For Zeeland, 20,698 out of 24,977 men from the 1812-1862 cohort

who died after their 50th birthday were studied. Occupational information is retrieved from marriage, death, and birth certificates of ego and his children. Unmarried men were excluded from the analyses, as only their occupation at death was known. Data needed to be converted to either an American or European coding standard in order to compare the effects of socioeconomic status in the Utah and Zeeland. The historical occupations in Utah were coded using the U.S. Bureau of the Census' 1950 standard, hereafter Occ1950, whereas

• ccupational data from Zeeland was coded into the HISCO standard. The Occ1950 can be converted

into NPBOSS, OCCSCORE, PRESGL, and SEI with crosswalks from IPUMS (IPUMS, 2017a, 2017b, 2017c, 2017d), while the HISCO can be recoded into HISCLASS, SOCPO and HISCAM with crosswalks from the IISH (K. Mandemakers et al., 2013). However, conversions from HISCO to NPBOSS, OCCSCORE, PRESGL, and SEI or from Occ1950 to HISCLASS, SOCPO and HISCAM are not possible. A crosswalk between Occ1950 and HISCO was deemed the optimal bases from which to test the American measurements of SES on European data and, vice versa, test the European measurement

• f SES and American data. For the conversion between HISCO and Occ1950 269 Occ1950
• ccupational categories and 1,675 HISCO occupational categories needed to be matched. The

conversion from Occ1950 to HISCO has partly been made by the North Atlantic Population Project (NAPP) that recoded Occ1950 into intermediate HISCO, a variation on HISCO also known as

9 OCCHISCO or NAPPHISCO. However, this coding system sometimes differs drastically from HISCO as intermediate HISCO was intended to replace both HISCO and Occ1950. Therefore, a new crosswalk had to be built. Occupational categories were linked between HISCO and Occ1950 on the basis of the underlying

• ccupations. Both HISCO and Occ1950 consist of multiple layers of occupational groups. HISCO is

divided in 7 major, 76 minor, 296 unit, and 1,675 micro groups, which roughly correspond with: social classes, sectors, occupational groups, and occupational subgroups. Occ1950 on the other hand is divided in 10 social classes and 269 occupational groups. HISCO’s micro groups and Occ1950’s

• ccupational subgroups are based on a well-documented number of occupations, which can easily be

compared and matched between both occupational coding schemes. In the translation from HISCO to Occ1950 1,675 occupational categories were collapsed into 229 unique Occ1950 occupational

• groups. Although 40 occupational groups in Occ1950 could not be retrieved from HISCO, all
• ccupations were successfully attributed to the right social class. Vice versa, 269 occupational groups

in Occ1950 were recoded into 227 unique HISCO micro groups. These 227 codes are well-spread over the different branches of the HISCO tree, as they cover practically all unit groups. Eight occupational scales are modeled for both Utah and Zeeland: Occ1950, HISCLASS, SOCPO, HISCAM, NPBOSS, OCCSCORE, PRESGL, and SEI. Table 1 shows an overview of these measurement

• schemes. Occ1950, HISCLASS, and SOCPO are categorical scales. Occ1950 and HISCLASS use inversed

scores for socioeconomic status, so that the lowest score is assigned to the highest social class. Occ1950 contains ten categories, ranging from 0 to 9. HISCLASS contains 13 categories, although it is

• ften collapsed into five categories. SOCPO contains six categories, ranging from 1 to 5, as category

four contains both farmers and the middle class, which are usually studied separately. The correlations in table 2 show that HISCLASS, Occ1950, and PRESGL strongly resemble one another in both Utah and Zeeland, whereas SOCPO is more related to the linear measurements of SES. HISCAM, NPBOSS, OCCSCORE, PRESGL, and SEI are linear measurements of SES that all apply different metric scales. HISCAM and SEI assign a SES-score on a 100-point scale, NPBOSS and PRESGL

• n a 1000-scale score, whereas OCCSCORE estimates the medium income in hundreds of U.S. 1950
• dollars. Furthermore, HISCAM and PRESGL scores do not use the whole metric scale, as HISCAM

scores range from 37 to 100, and PRESGL scores from 93-815. To be able to compare the outcomes

• f the different SES scales, all HISCAM, NPBOSS, OCCSCORE, PRESGL, and SEI scores have been

transformed to range from 0 to 1. Table 2 shows high correlations between the five different linear scales. Table 1: Overview of socioeconomic measurement schemes Base Abbreviation Name Range Ref Occupations HISCO Historical International Standard Coding of Occupations 0-7 (Van Leeuwen, Maas, & Miles, 2002, 2004) HISCO HISCLASS HISCLASS 1-13 (Van Leeuwen & Maas, 2011) HISCO SOCPO Social Power 1-5 (Van de Putte & Miles, 2005) HISCO HISCAM Historical CAMSIS 37-100 (Lambert, Zijdeman, Van Leeuwen, Maas, & Prandy, 2013) Occupations Occ1950 U.S. Bureau of the Census’ 1950 standard 0-9 (U.S. Bureau of the Census, 1950)

10 Occ1950 NPBOSS Nam-Powers-Boyd Occupational Status Score 0-1000 (Nam & Boyd, 2004; Nam & Powers, 1983) Occ1950 OCCSCORE Occupational Income Score 0-80 (IPUMS, 2017b) Occ1950 PRESGL Siegel Prestige Score 93-815 (Siegel, 1971) Occ1950 SEI Duncan Socioeconomic Index 0-100 (Duncan, 1961) Table 2: Correlations between SES measurements Utah and Zeeland Occ1950* HISCLASS* SOCPO HISCAM NPBOSS OCCSCORE PRESGL SEI Occ1950* .97

• .05
• .54
• .34
• .28
• .84
• .52

Zeeland HISCLASS* .81

• .07
• .52
• .28
• .23
• .82
• .48

SOCPO .61 .49 .67 .83 .77 .54 .77 HISCAM

• .03

.04 .49 .69 .58 .72 .75 NPBOSS .01 .16 .76 .63 .94 .70 .91 OCCSCORE -.01 .10 .66 .63 .91 .65 .87 PRESGL

• .56
• .49

.40 .39 .64 .67 .80 SEI

• .13
• .03

.74 .65 .92 .87 .75 Utah Table 3: Descriptive table Utah Zeeland mean se mean se N 21,386 Birth cohort 1860-1890 1812-1862 LE50 73.22 10.57 Migrated 46.6% Married 100% % second marriages 15.0% Age at first marriage 28.63 6.70 Age at first birth 28.89 5.78 Age at last birth N children Birth interval Father long-lived Mother long-lived Father died before 50 Mother died before 50 Age father at birth Age mother at birth N siblings Twin Analysis To study the impact of socioeconomic status on mortality after age 50 on the Utah 1860-1890 cohort and the Zeeland 1812-1862 cohort, Cox proportional hazard models were estimated. For both Utah and Zeeland, models were estimated with the coxme package (Therneau, 2012)in base R 3.3.2(R Core

11 Team, 2016). Besides information on ego’s socioeconomic status, the models controlled for father’s socioeconomic status, ego’s year of birth, and frailty among siblings. We furthermore included effects that could affect individual survival before the age of 50: early parental death, parental longevity, migration, being a twin, number of siblings, the birth interval to the former sibling, parental age at conception, and ego’s own fertility. In our models, farmers were used as a reference category, as they are a sizable group in society, are expected to outlive their peers, and are coded similarly between the different scales on socioeconomic status. Some categories of SES as well as some SES-scales were non-proportional, because effects of SES did not grow exponentially at higher ages. Nevertheless, we choose not to recode the available categories, since non-proportionality did not cause false positives or crossover

• effects. Moreover, combining categories would have disturbed the theoretical frameworks that we

are testing. Further inspection of the model showed misspecifications in the Cox model existed, but that they were rooted in temporal and rural-urban differences. Therefore, we present and discuss separate models in which we control for temporal and spatial effects. SES Effects of socioeconomic status on the mortality risk in later life strongly depend on the population and applied social class scheme. Both in Utah and Zeeland, all linear measures of socioeconomic status indicate a linear relationship between socioeconomic status and mortality risk in later life. HISCAM, Nam-Powers, OCCSCORE, PRESGL, and SEI indicate that the upper strata in Utah lived

• longer. For Zeeland, however, only PRESGL finds that a higher social standing is associated with

decreased mortality, whereas HISCAM, NPBOSS, OCCSCORE, and SEI yield the opposite effect.Linear measurements of SES indicate that the relationship between socioeconomic status on the morality risk in later life strongly differed between Utah and Zeeland. Categorical measurements of socioeconomic status show similar results for Utah. Occ1950, HISCAM, and SOCPO all show a more or less linear relationship between SES and mortality. The relationship is most linear for Occ1950 where mortality is lowest for professionals and farmers, and highest for operatives and laborers. At the bottom, however, class differences are less linear as

• peratives, farm laborers, and laborers show a much higher mortality than craftsmen and service
• workers. HISCLASS also finds a linear relationship for white collar professions (managers,

professionals, and clerical and sales personnel), and places lower skilled and unskilled farm workers and general workers at the bottom. But, just like Occ1950, HISCLASS has problems to stratify manual

• labor. SOCPO, on the other hand, does a good job in discerning the mortality differences between

unskilled, semiskilled and skilled workers, but does not find differences in mortality between the middle class, farmers, and the elite. Together these three measures show that mortalitywas gradually stratified among the skilled and white collar laborers, but that the rest of society (lower-skilled, unskilled, and farm laborers) showed highly elevated levels of mortality. Results are rather different for Zeeland. None of the three categorical measures of SES shows any evidence of a linear relationship between SES and mortality in Zeeland. Occ1950, HISCLASS, and SOCPO all find that mortality was lowest among farmers, but Occ1950 and HISCLASS also identify

• ther occupational groups with a lower mortality risk. According to Occ1950, craftsmen and farm

laborers outlived their peers, while HISCLASS identifies lower skilled farm workers and lower managers/proprietors as favorable categories. Craftsmen (medium and lower skilled workers) also perform well on HISCLASS, whereas lower managers (managers and proprietors) also perform well

12

• n the Occ1950. Besides farming, skilled labor and leading a small business thus seemed beneficial to

survival in Zeeland. Temporal controls Testing the models for cohort effects, shows some interesting results. For Utah, the linear effects of SES on mortality after age 50 are robust. However, the categorical measurements monitor some important changes over time. Occ1950 shows that the mortality risk for farm workers increasedto the level of operatives and laborers, whereas the service personal started to live longer and slowly

• vertook the craftsmen in terms of survival. The declining position of agricultural workers is also

picked up by HISCLASS, showing that the difference between the bottom of the social pyramid (farm, lower and unskilled workers) and the rest of the social pyramid becomes more pronounced over

• time. None of these trends are picked up by SOCPO. Nevertheless, Occ1950, HISCLASS and SOCPO

still show a more or less linear relationship between socioeconomic status and mortality after age 50 for all periods under study. This relationship is most linear for Occ1950, less pronounced for HISCLASS, and absent for the middle and upper class for SOCPO. Differences are more pronounced for Zeeland. No relationship between socioeconomic status and mortality risk after age 50 was found in the first three decades of cohorts under observation by HISCAM, NPBOSS, OCCSCORE, and SEI, whereas PRESGL found that a higher socioeconomic status corresponded with lower mortality. In the taller two decades, however, four out of five – HISCAM, NPBOSS, OCCSCORE, and SEI – these measures found that a higher socioeconomic status corresponded with higher mortality. At a first glance, Occ1950 and HISCLASS seem to find the same results as HISCAM, NPBOSS, OCCSCORE, and SEI. However, mortality risks did not change for most socioeconomic classes. According to Occ1950, mortality increased for the elite and operatives, while at the bottom mortality lowered for service workers and laborers. HISCLASS shows that conditions for lower managers and laborers improved, but worsened for lower professionals / higher clerical and sales workers. SOCPO only reports elevated mortality for semi-skilled workers.The linear relationship between socioeconomic status and mortality is thus based onshifting mortality risks in two social groups: lower professionals at the top of the social pyramid, and laborers at the bottom. Rural-urban Whether individuals lived in a rural or urban environment also caused some minor, but important

• differences. HISCAM, NPBOSS, OCCSCORE, PRESGL, and SEI indicate that the relationship between

socioeconomic status and mortality risks after age 50 was stronger in rural environments than in urban Ogden, Provo, or Salt Lake City.The main difference between the Utahn countryside and cities is the stratification at the bottom of the social pyramid. In the towns as well as the cities, mortality risk decrease gradually for the white collar professions. Mortality risks are lowest among the elite and farmers, slightly higher for managers/proprietors and merchants, more increased for craftsmen, and even higher for operatives. However, craftsmen/skilled laborers live much longer on the countryside than in the city. Service workers also fare better in a rural environment, as their mortality risk is comparable to that of farmers, rather than operatives. Moreover, unskilled laborers and agricultural laborers are also worse off in the city, as the mortality between lower skilled, agricultural, and unskilled labor gradually increases in urban Utah, whereas it is leveled for the rest of Utah. For Zeeland, HISCAM, NPBOSS, OCCSCORE, and SEI report little differences between the cities of Goes, Middelburg, and Vlissingen and the rest of Zeeland. According to these measurements, a

13 higher socioeconomic corresponds with higher mortality, although the effect is insignificant in the smaller urban sample. The estimate of PRESGL differs between the two environments. In the towns, PRESGL indicates that a higher socioeconomic status leads to a significantly lower mortality risk, whereas the relationship is positive – although not significant – for the three biggest cities in Zeeland. Occ1950 and SOCPO the estimated mortality risks give little evidence for rural-urban differences. This makes it remarkable that the effect of PRESGL on mortality differs between the two

• environments. Apparently, even the smallest shifts in the labor market affect the estimates of linear

measures of socioeconomic status. Conclusion This paper discussed whether socioeconomic stratification of mortality after age 50 also existed for 19th-century cohorts. Earlier studies have shown that modern-day social inequalities in survival were already present in Quebec and on the Utahn frontier, but that an inversed relationship between socioeconomic status and survival in later life could not be established for European populations. However, it was unclear whether these different results were caused by methodological decisions or actual historical differences. By comparing eight schemata of socioeconomic status on the Utah 1860-1890 cohort to the Zeeland 1812-1862 cohort, we were able to determine that class-based inequalities in later-life survival were strong for Utah and weak for Zeeland. Moreover, socioeconomic inequalities in survival followed a different pattern in the 19th-century commercial- agricultural society. Socioeconomic measurement schemata had a significant impact on our results. For Utah, continuous and categorical scales show the same general pattern between socioeconomic status and survival in later life. But, categorical schemata are better equipped to pick up changes over time, differences between regions, and divergences from the linear trend. In Zeeland, continuous measurements perform poorly, as they produce an artificial, positive relationship between socioeconomic status and mortality risks in later life and fail to identify existing socioeconomic inequalities in later-life survival. Therefore, future studies should refrain from using continuous measurements of socioeconomic status. Of all schemata, OCC1950 and HISCLASS give the most reliable and robust outcomes, with OCC1950 slightly outperforming HISCLASS as it is conceptually clearer on the socioeconomic position of farmers, better distinguishes different forms of manual labor, and has a more feasible number of categories. Methodological decisions had little impact on the results for Utah. Mortality risks differed gradually between the white collar classes, service workers, and craftsmen, and were significantly elevated for operatives, agricultural and general laborers. Most of these effects were stable over time, even though the Utahn labor market diversified considerably between 1880 and 1940. However, mortality risks increased slightly for agricultural and general laborers, while they decreased for service workers. Mortality risks differed more between rural and urban Utah. In Ogden, Provo, and Salt Lake City mortality risks increased gradually for all social classes, whereas in rural Utah

• peratives, agricultural laborers, and general laborers had significantly elevated mortality risks and

service workers resembled white collar workers. As such, higher socioeconomic status went hand in hand with lower mortality risks in later life in Utah, especially for the urban area. Class-based inequalities in later-life survival in Zeeland were more in line with earlier studies on European commercial-agricultural populations. Mortality risks after age 50 were elevated for the elite and lowest for farmers. Analyses further show lower mortality risks for shippers, craftsmen, and agricultural laborers, whereas peddlers had higher mortality risks. The differences between these

14 groups are rather small and would not have been picked up with a smaller sample. Therefore, it is not surprising that class-based inequalities in later-life survival do not significantly differ between cohorts or between rural and urban settlements. Stratification of later-life mortality was only marginally present in Zeeland and did not correlated with increased social standing. These findings for Utah and Zeeland give three important insights in the 19th-century stratification

• f health. First, the elite had a lower later-life expectancy in the latter half of the 19th and first half of

the 20th century due to obesity, smoking, and drinking. That the wealthiest in Utah lived longer than their fellow statesmen is a surprising finding, as other studies report a negative relationship between socioeconomic status and survival after age 50 (Edvinsson & Broström, 2012; Ferrie, 2003; Gagnon et al., 2011; Schenk & van Poppel, 2011). Our hypothesis was that the elite would suffer from “diseases

• f affluence” due to obesity and smoking and drinking behavior. It might be that there was simply

not enough food for overconsumption on the frontier, or that healthy lifestyle were enforced by the Church of Latter-Day Saints. To test the latter hypothesis we compared the mortality risks of Mormon and non-Mormon professionals, as practicing Mormons are known to outlive non-church members (Lindahl-Jacobsen et al., 2013). However, we could not find any difference in the mortality risks of active, inactive and non-Mormon professionals. Most likely, life in Utah left little room for conspicuous overconsumption and the absence of a survival penalty for the Utahn elite should be seen as a local divergence from a global trend. Second, there seems to be little evidence of a strong socioeconomic differences in survival in commercial-agricultural societies. Results from Zeeland are in line with results from other European

• populations. Minor differences existed, and indicate that farmers, farm laborers, craftsmen, and

shippers fared well, whereas peddlers fared worse. Similar findings have been widely reported for farmers (Edvinsson & Broström, 2012; Ferrie, 2003; Gagnon et al., 2009; Schenk & van Poppel, 2011; Temby & Smith, 2014) and small-scale entrepreneurs also had higher mortality in Sundsvall (Edvinsson & Broström, 2012), but the effects for farm laborers, craftsmen, and shippers are new. The positive for farm laborers is also radically different for Utah where farm workers were at the bottom of the social pyramid. We attribute these findings to the importance of food security. In Utah, farming was a family business and only the poorest remained agricultural workers, whereas agricultural labor it was a form of skilled labor in Zeeland. Farm laborers, farmers, craftsmen, and shippers were the most food secure in commercial-agricultural populations, which gave them a marginal advantage in later-life survival. Third, Utah was a forerunner in terms of socioeconomic stratification. Increased social standing came with increased survival in later life, a pattern that is rarely found in other locale. It is doubtful that the gains in survival were attributable to gains in food security and vitamin intake, as results from Zeeland show that these gains were marginal at best. A possible explanation is the industrialization of the Utahn society. In Utah, the share of men who worked their entire life as either an semi- or unschooled laborer was rather low. The possibilities of upward mobility might have created class differences in survival. The lack of stratification in blue collar professions on the Utahn countryside shows that such a pattern could only exist if the labor markets were open and

• diversified. Therefore, the inversed pattern between socioeconomic status and mortality that we find

today might be the result of health selection effects. Combined, these three findings show that that the nowadays omnipresent differentials in survival are a product of our time. Although stratification of wealth and later-life survival occurred in pre- industrial populations, they were not necessarily distributed among class lines. Moreover, contemporary stratification of mortality might be more strongly rooted in selection effects than is

15 commonly assumed. To better understand contemporary and 19th-century differentials in survival, two question on the nature of social stratification need to be answered. First, when did the contemporary, inversed pattern between socioeconomic status and mortality start to appear, and, second, why is the effect of socioeconomic status on mortality so axiomatic today if it is not inevitable. Literature Alter, G., Dribe, M., & Van Poppel, F. W. A. (2007). Widowhood, family size, and postreproductive mortality: A comparative analysis of three populations in nineteenth-century Europe. Demography, 44(4), 785–806. https://doi.org/10.1353/dem.2007.0037 Alter, G., Neven, M., & Oris, M. (2004). Height, wealth and longevity in XIXth century east Belgium. Annales de Démographie Historique, 2(108), 19–37. https://doi.org/10.3917/adh.108.0019 Antonovksy, A. (1967). Social class, life expectancy and overall mortality. The Milbank Memorial Fund Quarterly, 45(2), 31–73. Bean, L. L., Mineau, G. P., & Anderton, D. L. (1990). Fertility change on the American frontier: Adaptation and innovation. Berkeley: University of California Press. Beekink, E., & Kok, J. (2017). Temporary and lasting effects of childhood deprivation on male stature. Late adolescent stature and catch-up growth in Woerden (The Netherlands) in the first half of the nineteenth century. History

• f

the Family, 22(2–3), 196–213. https://doi.org/10.1080/1081602X.2016.1212722 Bengtsson, T., & Dribe, M. (2011). The late emergence of socioeconomic mortality differentials: A micro-level study of adult mortality in southern Sweden 1815–1968. Explorations in Economic History, 48(3), 389–400. https://doi.org/10.1016/j.eeh.2011.05.005 Bengtsson, T., & van Poppel, F. (2011). Socioeconomic inequalities in death from past to present: An introduction. Explorations in Economic History, 48(3), 343–356. https://doi.org/10.1016/j.eeh.2011.05.004 Bobak, M., Pikhart, H., Rose, R., Hertzman, C., & Marmot, M. (2000). Socioeconomic factors, material inequalities, and perceived control in self-rated health: Cross-sectional data from seven post- communist countries. Social Science and Medicine, 51(9), 1343–1350. https://doi.org/10.1016/S0277-9536(00)00096-4 Brooshooft, P. (1897). Officieele en feitelijke waarheid: Bijdrage tot de kennis van onze

• arbeiderstoestanden. Den Haag: Gebr. Belinfante.

Chen, F., Yang, Y., & Liu, G. (2010). Social change and socioeconomic disparities in health over the life course in China: A cohort analysis. American Sociological Review, 75(1), 126–150. https://doi.org/10.1177/0003122409359165 Christensen, K., & Vaupel, J. W. (1996). Determinants of longevity: Genetic, environmental and medical factors. Journal of International Medicine, 240, 333–341. Davis, K. (1945). The world demographic transition. Annals of the American Academy of Political and Social Science, 237(1), 1–11. Donrovich, R., Puschmann, P., & Matthijs, K. (2014). Rivalry, solidarity, and longevity among siblings: A life course approach to the impact of sibship composition and birth order on later life mortality risk, Antwerp (1846–1920). Demographic Research, 31, 1167–1198. https://doi.org/10.4054/DemRes.2014.31.38 Duncan, O. D. (1961). A socioeconomic index for all occupations. In A. J. Reiss (Ed.), Occupations and social status. Glencoe, Illinois: Free Press of Glencoe. Dyson, T. (2010). Population and development: The demographic transition. London: Zed Books. Edvinsson, S., & Broström, G. (2012). Old age, health, and social inequality: Exploring the social patterns of mortality in 19th century northern Sweden. Demographic Research, 26, 633–660. https://doi.org/10.4054/DemRes.2012.26.23 Elo, I. T. (2009). Social Class Differentials in Health and Mortality: Patterns and Explanations in Comparative Perspective. Annual Review

• f

Sociology, 35(1), 553–572.

16 https://doi.org/10.1146/annurev-soc-070308-115929 Ferrie, J. P. (2003). The Rich and the Dead: Socioeconomic Status and Mortality in the United States, 1850-1860. In D. L. Costa (Ed.), Health and Labor Force Participation over the Life Cycle: Evidence from the Past (pp. 11–50). Chicago: University of Chicago Press. Retrieved from http://ideas.repec.org/h/nbr/nberch/9627.html Floud, R., Fogel, R. W., Harris, B., & Hong, S. C. (2011). The changing body: Health, nutrition, and human development in the Western World since 1700. Cambridge: Cambridge University Press. Gagnon, A., Smith, K. R., Tremblay, M., Vézina, H., Paré, P.-P., & Desjardins, B. (2009). Is there a trade-off between fertility and longevity? A comparative study of women from three large historical databases accounting for mortality selection. American Journal of Human Biology, 21(4), 533–540. https://doi.org/10.1002/ajhb.20893 Gagnon, A., Tremblay, M., Vézina, H., & Seabrook, J. A. (2011). Once were farmers: Occupation, social mobility, and mortality during industrialization in Saguenay-Lac-Saint-Jean, Quebec 1840–1971. Explorations in Economic History, 48(3), 429–440. https://doi.org/10.1016/j.eeh.2011.05.008 Huisman, M., Kunst, A. E., & Mackenbach, J. P. (2003). Socioeconomic inequalities in morbidity among the elderly; A European overview. Social Science and Medicine, 57(5), 861–873. https://doi.org/10.1016/S0277-9536(02)00454-9

• IPUMS. (2017a). NPBOSS50: Nam-Powers-Boyd occupational status score, 1950 basis. Retrieved from

https://usa.ipums.org/usa-action/variables/NPBOSS50#description_section

• IPUMS. (2017b). OCCSCORE: Occupational income score. Retrieved from https://usa.ipums.org/usa-

action/variables/OCCSCORE#description_section IPUMS. (2017c). PRESGL: Occupational prestige index, Siegel. Retrieved from https://usa.ipums.org/usa-action/variables/PRESGL#description_section

• IPUMS. (2017d). SEI: Duncan socioeconomic score. Retrieved from https://usa.ipums.org/usa-

action/variables/SEI#description_section Janssens, A. (2014). Labouring Lives: Women, Work and the Demographic Transition in the Netherlands, 1880-1960 (Vol. 18). Frankfurt am Main: Peter Lang. https://doi.org/10.3726/978- 3-0352-0273-1 Kannisto, V. (2001). Mode and dispersion of the length of life. Population: An English Selection, 13(1), 159–171. https://doi.org/10.2307/3030264 Kunst, A. E., Bos, V., Andersen, O., Cardano, M., Costa, G., Harding, S., … Mackenbach, J. P. (2004). Monitoring of trends in socioeconomic inequalities in mortality: Experiences from a European project. Demographic Research, 10(SUPPL. 2), 229–254. https://doi.org/10.4054/DemRes.2004.S2.9 Kunst, A. E., Looman, C. W. N., & Mackenbach, J. P. (1990). Socio-economic mortality differences in the Netherlands in 1950-1984: A regional study of cause-specific mortality. Social Science and Medicine, 31(2), 141–152. https://doi.org/10.1016/0277-9536(90)90055-W Lambert, P. S., Zijdeman, R. L., Van Leeuwen, M. H. D., Maas, I., & Prandy, K. (2013). The construction

• f HISCAM: A stratification scale based on social interactions for historical comparative
• research. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 46(2), 77–
• 89. https://doi.org/10.1080/01615440.2012.715569

Lindahl-Jacobsen, R., Hanson, H. A., Oksuzyan, A., Mineau, G. P., Christensen, K., & Smith, K. R. (2013). The male-female health-survival paradox and sex differences in cohort life expectancy in Utah, Denmark, and Sweden 1850-1910. Annals of Epidemiology, 23(4), 161–166. https://doi.org/10.1016/j.annepidem.2013.02.001 Mackenbach, J. P., Stirbu, I., Roskam, A.-J. R., Schaap, M. M., Menvielle, G., Leinsalu, M., & Kunst, A.

• E. (2008). Socioeconomic Inequalities in Health in 22 European Countries. New England Journal
• f Medicine, 358(23), 2468–2481. https://doi.org/10.1056/NEJMsa0707519

Mandemakers, K., & Laan, F. (2017). LINKS dataset Genes Germs and Resources, WieWasWie Zeeland, Civil Certificates, version 2017.01 [Data file and code book]. Amsterdam: IISH. Mandemakers, K., Muurling, S., Maas, I., Van de Putte, B., Zijdeman, R. L., Lambert, P., Van Leeuwen,

• M. H. D., … Miles, A. (2013). HSN standardized, HISCO-coded and classified occupational titles,

17 release 2013.01. IISH. Retrieved from https://socialhistory.org/en/hsn/hsn-standardized-hisco- coded-and-classified-occupational-titles-release-201301?language=en Marmot, M. G., Adelstein, A. M., Robinson, N., & Rose, G. A. (1978). Changing social-class distribution

• f

heart disease. British Medical Journal, 2(6145), 1109–1112. https://doi.org/10.1136/bmj.2.6154.1785-a Mazzoni, S., Breschi, M., Manfredini, M., Pozzi, L., & Ruiu, G. (2017). The relationship between family characteristics and height in Sardinia at the turn of the twentieth century. History of the Family, 22(2–3), 291–309. https://doi.org/10.1080/1081602X.2016.1230509 McKeown, T. (1976). The modern rise of population. London: Edward Arnold Publishers. Mourits, R. (2017). Historical Life Course Studies. Historical Life Course Studies, 1, 1–19. Retrieved from http://www.ehps-net.eu/article/later-life-mortality-and-longevity-late-18th-and-19th- century-cohorts-where-are-we-now-and/hislives_vol_4_27.01.17.pdf Nam, C. B., & Boyd, M. (2004). Occupational status in 2000: Over a century of census-based measurement. Population Research and Policy Review, 23(4), 327–358. https://doi.org/10.1023/B:POPU.0000040045.51228.34 Nam, C. B., & Powers, M. G. (1983). The socioeconomic approach to status measurement: With a guide to occupational and socioeconomic status scores. Houston, Texax: Cap and Gown Press. Notestein, F. W. (1945). Population: The long view. In T. W. Schultz (Ed.), Food for the World (pp. 36– 57). Chicago: University of Chicago Press. Omran, A. R. (1971). The epidemiologic transition: A theory of the epidemiology of population change. The Milbank Quarterly, 49(4), 509–538. https://doi.org/10.1111/j.1468- 0009.2005.00398.x Preston, S. H. (1976). Mortality Patterns in National Populations: With Special Reference to Recorded Causes of Death (Studies in population). (H. H. Winsborough, Ed.). New York: Academic Press. Retrieved from http://www.sciencedirect.com/science/book/9780125644501 Priester, P. (1998). Geschiedenis van de Zeeuwse landbouw circa 1600-1910. Wageningen: Landbouwuniversiteit Wageningen. R Core Team. (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.r-project.org/ Ramon-Muñoz, R., & Ramon-Muñoz, J. M. (2017). Sibship size and the biological standard of living in industrial Catalonia, c.1860–c.1920: a case study. History of the Family, 22(2–3), 333–363. https://doi.org/10.1080/1081602X.2016.1224727 Razzell, P., & Spence, C. (2006). The hazards of wealth: Adult mortality in pre-twentieth-century

• England. Social History of Medicine, 19(3), 381–405. https://doi.org/10.1093/shm/hkl048

Robine, J.-M., & Paccaud, F. (2005). Nonagenarians and centenarians in Switzerland, 1860-2001: a demographic analysis. Journal of Epidemiology and Community Health, 59(1918), 31–37. https://doi.org/10.1136/jech.2003.018663 Rotberg, R. I., & Rabb, T. K. (1985). Hunger and history: The impact of changing food production and consumption patterns on society. Cambridge: Cambridge University Press. Schenk, N., & van Poppel, F. (2011). Social class, social mobility and mortality in the Netherlands, 1850–2004. Explorations in Economic History, 48(3), 401–417. https://doi.org/10.1016/j.eeh.2011.05.010 Schrijvers, C. T. M., Stronks, K., Van de Mheen, H. D., & Mackenbach, J. P. (1999). Explaining educational differences in mortality: The role of behavioral and material factors. American Journal of Public Health, 89(4), 535–540. https://doi.org/10.2105/AJPH.89.4.535 Siegel, P. M. (1971). Prestige in the American occupational structure. Chicago: University of Chicago. Smith, K. R., Mineau, G. P., Garibotti, G., & Kerber, R. (2009). Effects of childhood and middle- adulthood family conditions on later-life mortality: evidence from the Utah Population Database, 1850-2002. Social Science & Medicine, 68(9), 1649–1658. https://doi.org/10.1016/j.socscimed.2009.02.010 Temby, O. F., & Smith, K. R. (2014). The association between adult mortality risk and family history of longevity: The moderating effects of socioeconomic status. Journal of Biosocial Science, 46(6),

18 703–716. https://doi.org/10.1017/S0021932013000515 Therneau, T. M. (2012). coxme: Mixed Effects Cox Models. R package version 2.2-5. Retrieved from https://cran.r-project.org/package=coxme U.S. Bureau of the Census. (1950). Alphabetical index of occupations and industries: 1950. Washington, D.C.: U.S. Bureau of the Census. Van de Putte, B., & Miles, A. (2005). A social classification scheme for historical occupational data. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 38(2), 61–92. https://doi.org/10.3200/HMTS.38.2.61-94 Van Leeuwen, M. H. D., & Maas, I. (2007). Economische specialisering en veranderende sociale verhoudingen in de 19e en 20e eeuw: Een studie op basis van de Nederlandse volkstellingen en

• huwelijksakten. In Twee eeuwen Nederland geteld (pp. 181–206). Den Haag: DANS - Data

Archiving and Networked Services. Van Leeuwen, M. H. D., & Maas, I. (2011). HISCLASS: A historical social class scheme. Leuven: Leuven University Press. Van Leeuwen, M. H. D., Maas, I., & Miles, A. (2002). HISCO: Historical International Standard Classification of Occupations. Leuven: Leuven University Press. Van Leeuwen, M. H. D., Maas, I., & Miles, A. (2004). Creating a historical international standard classification of occupations: An exercise in multinational interdisciplinary cooperation. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 37(4), 186–197. https://doi.org/10.3200/HMTS.37.4.186-197 Van Zanden, J. L., & Van Riel, A. (2000). Nederland 1780-1914: Staat instituties en economische

• ntwikkeling. Amsterdam: Balans.

Walhout, E., & van Poppel, F. (2003). De vermelding des beroeps: eene ijdele formaliteit? Twee eeuwen vrouwelijke beroepsarbeid in Nederlandse huwelijksakten. Tijdschrift Voor Sociale Geschiedenis, 23(3), 301–332. Wilmoth, J. R. (2000). Demography of longevity: Past, present, and future trends. Experimental Gerontology, 35(9–10), 1111–1129. https://doi.org/10.1016/S0531-5565(00)00194-7 Zimmer, Z., Hanson, H. A., & Smith, K. (2016a). Childhood socioeconomic status, adult socioeconomic status, and old-age health trajectories. Demographic Research, 34(February), 285–320. https://doi.org/10.4054/DemRes.2016.34.10 Zimmer, Z., Hanson, H. A., & Smith, K. R. (2016b). Offspring socioeconomic status and parent mortality within a historical population. Demography, 53(5), 1583–1603. https://doi.org/10.1007/s13524-016-0502-x

19 Appendix Occ1950 Utah Zeeland

• 0. professionals

4,436 9.58% 405 1.89%

• 1. farmers

21,969 47.46% 5,470 25.58% 2. managers & proprietors 4,958 10.71% 3,086 14.43%

• 3. clerical

1,723 3.72% 404 1.89%

• 4. sales workers

1,220 2.64% 197 0.92%

• 5. craftsmen

5,506 11.90% 3,700 17.30%

• 6. operatives

2,296 4.96% 962 4.50%

• 7. service worker

692 1.50% 434 2.03%

• 8. farm laborer

1,980 4.28% 4,394 20.55%

• 9. laborer

1,507 3.26% 2,334 10.91% HISCLASS Utah Zeeland 1a. higher manager 465 1.03% 710 3.32% 1b. higher professional 3,501 7.74% 250 1.17%

• 2a. lower manager

8,250 18.24% 946 4.42% 2b. lower professional, higher and middle clerks & salesmen 1,938 4.28% 2,583 12.08%

• 2c. lower clerks

and salesmen 3,398 7.51% 383 1.79%

• 2d. foremen

352 0.78% 71 0.33%

• 3a. medium skilled

4,865 10.75% 3,001 14.03%

• 3b. lower skilled

2,726 6.03% 1,640 7.67% 4. farmers and fishermen 16,005 35.38% 4,635 21.68%

• 5a. lower skilled

farm workers 1,099 2.43% 260 1.22% 5b. unskilled workers 2,086 4.61% 686 3.21%

• 5c. unskilled farm

workers 552 1.22% 4,251 19.88% 5d. unskilled workers, not specified

• 1,967

9.20% SOCPO Utah Zeeland

• 1. unskilled

4,635 10.25% 6,916 32.34%

• 2. semi-skilled

2,219 4.91% 1,760 8.23%

• 3. skilled

4,928 10.89% 3,029 14.15%

• 4a. middle class

9,181 20.30% 4,683 21.90%

• 4b. farmers

20,440 45.18% 4,735 22.14%

• 5. elite

3,834 8.48% 263 1.23%