Working Paper ICP Please do not cite without permission from the - - PDF document

1

Working Paper ICP Please do not cite without permission from the author

Session: Mortality patterns in the past Monday 30 October 8:30-10:30 Tim Riswick This working paper is part of the PhD-project ‘Between Rivalry and Support: Differences in Infant and Child Mortality Chances of Brothers and Sisters in the Netherlands (1863-1910) and Taiwan (1906- 1946)’ and only includes the study of sibling effects within the Netherlands. During the presentation on the International Population Conference some preliminary results on Taiwan will also be presented to be able to show some similarities and differences in sibling effects between these two societies. Abstract ‘Testing the Conditional Resource Dilution Hypothesis: The Impact of Sibling Size and Composition on Infant and Child Mortality in the Netherlands, 1863-1910’ Child survival depends on the allocation of resources within the household. The size and composition

• f the sibling set influence parental division of resources and can in turn affect survival chances. In spite
• f recent research advances on sibling effects, studies often use the resource dilution hypothesis that

neglects the specific historical context which shapes household structure and organization. This study therefore focuses specific on the historical context by examining sibling effects on infant and child mortality in three regions in the Netherlands in the period 1863-1910. It does so by using the conditional resource dilution model as a conceptual framework, which incorporates economic conditions, cultural codes and practices and family systems, and by taking changing household composition into account by using longitudinal data from the Historical Sample of the Netherlands, time-varying variables and Cox proportional-hazard models to study sibling effects. The results suggest that, next to the influence of socio-economic and biological determinants, the number and gender of siblings plays an important role for child mortality, but less for infant mortality. These effects are, however, gendered because the number of same-sex siblings mainly has an impact on boys’ survival chances. Moreover, when examining the three regions a negative influence of the number of sisters is only found for child mortality chances for girls in the nuclear family region. These findings underscore the importance of looking at the specific historical context when trying to understand how sibling size and composition may have different effects of children’s mortality chances.

2

Testing the Conditional Resource Dilution Hypothesis: The Impact of Sibling Size and Composition on Infant and Child Mortality in the Netherlands, 1863-1910

Tim Riswick

• 1. Introduction

The position of a certain child within the household is considered to be essential because infants and young children are totally dependent on tangible and intangible resources from the household and its members (Bengtsson, Campbell, and Lee 2004).The focus of most research is mostly on the influence

• f the presence or absence of parents. These studies, for example, demonstrate the importance of the

mother for the survival of her children (Van Poppel 2000; Van Poppel and Gaalen 2008), while the importance of the father is much more debated and depends on household structure and historical context (Derosas and Oris 2002). Yet, most of the time households in pre-modern times did not only consist of a father and mother: other present (adult) kin could also be able to take over some roles of parents and

• ffer support or have a negative influence because of increased competition over household resources.

Overview studies looking into kin effects on mortality outcomes especially emphasize the positive effect

• f kin by concluding that all studies suggest that at least one relative - apart from the mother - is

beneficial for child survival in almost all populations. There is much variation in how relatives are helpful to the survival of children, but siblings are often overlooked and therefore rarely studied in research examining infant and child survival (Sear and Coall 2011; Sear and Mace 2008). This study therefore tries to answer the question under what circumstances, and through which mechanisms, are sibling size and composition actually beneficial or detrimental for child survival chances? Earlier research investigating the effect of sibship size on other demographic processes and

• utcomes explain the influence of siblings with the resource dilution model. According to this model

parental resources are finite and additional children dilute these resources. The amount of resources that can be assigned to each child in the household is, therefore, only dependent on both the amount of resources and the number of children (Downey 2001). Although several scholars have uncovered extensive empirical evidence which is consistent with the resource dilution model (Black, Devereux, and Salvanes 2005; Booth and Kee 2008; Carter et al. 2002; Downey 1995; Dribe, Campbell, and Van Bavel 2012), the number of empirical studies that challenge it is growing (Bras, Kok, and Mandemakers 2010; Chu, Xie, and Yu 2007; Li, Zhang, and Zhu 2008; Lu, Treiman, and Donald 2014; Marteleto and de Souza 2012; Shavit, Pierce, and Pierce 1991; Yu, Su, and Chiu 2012). These studies demonstrate that the theory is too simplistic because the parental couple is regarded to be the unit that decides on the distribution of resources. Other possible factors which may influence the household strategy are

3 therefore ignored. As a result, more and more scholars are advocating for a more contextual and flexible conceptual framework. One that is able to incorporate specific cultural codes and practices, economic conditions and institutions to determine how sibship size matters (Gibbs, Workman, and Downey 2016; Öberg 2017; Powell, Werum, and Steelman 2004). Moreover, it is also argued that gender regimes should be included because some form of gender discrimination exists in most societies (Das Gupta 1997; Kalmijn and van de Werfhorst 2016). Yet, to be able to investigate the ideas of the conditional or gendered resource dilution model a more comparative approach is needed, which is lacking in most studies at this moment. Given how little is known about the connection between sibship size and composition and survival chances, this study investigates how and why sibling effects affect infant and child mortality between the ages zero and five in the Netherlands in the period 1863-1910. On the one hand, examining how siblings influence survival chances is important because it sheds light on the most extreme inequalities within the household which eventually result in death. On the other hand, this study also contributes to the overall literature on sibling effects by (1) using the conditional resource dilution model as a conceptual framework, which incorporates economic conditions, cultural practices and family and gender systems, and by (2) using longitudinal data and time-varying variables which can take changing household composition throughout time into account. The latter is possible because of a sample from Historical Sample of the Netherlands and by using Cox proportional-hazard event history analysis. In sum, by studying the connection between sibling effects and survival chances of young children, this study advances the literature on both topics. It offers evidence for the influence of siblings on demographic processes and outcomes in general, and for mortality in particular. In the following section, the mechanisms and previous research findings based on the resource dilution hypothesis are used to formulate expectations about mechanisms by which possible sibling effects influenced infant and child mortality. Next, hypotheses originating from the theory and specific historical context are formulated. After the datasets and methods are introduced, the results of the descriptive statistics and Cox proportional-hazard models are discussed. The article concludes with a discussion on the interpretation of the influences of siblings on infant and child survival in the Netherlands and some suggestions for future research.

• 2. Background: incorporating regional variation and gender in the resource dilution model

The resource dilution hypothesis is the most used theory that addresses the way in which the number of siblings relates to the allocation of resources in the household. There are, however, some problems with the exact features of this hypothesis. A first point of critique is that the model argues that parental resources are finite, while this does not imply that the total amount and distribution of these resources is fixed. The amount and distribution of resources can also change over time (Downey 2001). Moreover, gender systems may play an important role in the exact allocation and quality of resources because boys

4 and girls are treated differently (Kalmijn and van de Werfhorst 2016). A second point of critique is that the resource dilution model is overemphasizing the primacy of parental resources. From this perspective the household is the unit which decides on the distribution of resources. Nevertheless, other actors can also play a significant role because they can add to the amount or division of resources within a

• household. Siblings, for example, could also serve as resources rather than as competitors because they

can generate additional income or provide care, as the concept of humans as cooperative breeders suggests (Hrdy 2011). In other words, the focus in the model is on nuclear family processes which downplays or ignores the impact of other kin and the larger community (Gibbs et al. 2016). A last point

• f critique is the way in which resources are perceived, because a distinction should be made between

base and surplus resources. Base resources, such as nutrition and care, are critical for survival, while surplus resources are not and are mostly considered as investments to enhance children’s life chances and long term human capital (Downey 2001). To reconcile most of these critiques the conditional resource dilution model is proposed. The advantage of this kind of model is that it is more flexible because it allows for contextual factors, while it still follows the general tenet of dilution of parental resources when sibship size increases. The degree to which children’s development is a product of factors other than parental investments can therefore also be included (Gibbs et al. 2016). One way to grasp the influence of these contextual factors is to borrow some elements from ecological systems theory. The dilution processes within the household can, in my view, also be defined as a microsystem, that is embedded within the larger context of how family and broader kinships are organized in a society (Bronfenbrenner 1979, 1986). These universal principles of family organization and relationships are anchored in regional family cultures, which are typically defined as so-called family or kinships systems (Therborn 2004; Todd 1985). This includes the “beliefs and norms, common practices, and associated sanctions through which kinship and the right and obligations of particular kin are defined” (Mason 2001: 160). Family systems in turn reflect the customary, normative manner in which family practices and household dynamics unfold (Mönkediek 2016; Skinner 1997). Additionally these systems are intertwined with gender regimes and may promote a certain gender order, as well as social order, causing inequalities within the household (Das Gupta 1997; Hilevych 2016). The most common elements to distinguish between principles of family

• rganization and structure are marriage, household and inheritance practices (Goody 1996; Das Gupta

1999; Reher 1998), while others have also emphasized the importance of parental power (Klep 2005; Wolf 2005). Thus, there are many arguments to assume that family systems, as being part of the contextual factors, frame people’s demographic behaviors by structuring people’s social relationships and their family experience. The deposition is therefore that how sibship size and composition matters depends on broader societal characteristics such as regional family systems. Most theories of, and research on, the mentioned aspects of the resource dilution model focus

• n cognitive ability and educational attainment. It is argued that resources may play a crucial role

5 because educational attainment is lower among children in larger families because they are likely to receive less parental attention, care and other resources (Black et al. 2005; Carter et al. 2002; Conley 2000; Downey 2001) It can, however, also be argued that sibship structures may also be associated with

• ther individual outcomes, such as health, insofar as they affect parents’ expectations and in turn the

specific amount of household resources towards a child. Especially, sibship characteristics are likely to influence individuals’ survival chances as an extra mouth to feed or hand to help could make a huge

• difference. Moreover, if the household is the only or primary source of base resources, than size and

structure of the sibling set may be important for survival chances too. Lastly, focusing on the relation between the resource dilution hypothesis and health offers new insights because it puts the focus on base resources instead of surplus resources as most studies focusing on children’s status attainment do. Despite these possibilities, scant previous research examines the association between sibship structures and health in early childhood in historical societies.1 A line of research, which proved to be fruitful during the last years, is the examination of the relationship between sibship size and height differences as an indicator of health. The argument is mainly based on the relation between sibship size, income and nutritional dilution, crowding and the higher chances of infection in larger families (Bailey, Hatton, and Inwood n.d.; Hatton 2016). Yet, de results are mixed: some studies find a significant negative effect of the number of siblings on the height of individuals (Keyser and Rossem 2016; Mazzoni et al. 2016; Myrskylä et al. 2013; Roberts and Warren 2016; Stradford, van Poppel, and Lumey 2016), but others find effects which are weak or disappear over time (Beekink and Kok 2016; Öberg 2015; Poulain et al. 2017; Ramon-Muñoz and Ramon-Muñoz 2016). The latter argue that the changing role of sibship size could be caused by fertility decline, the general improvement of standards of living, the development of the welfare state and improving health. Öberg (2015) already demonstrated this in his study of Southern Sweden where the strength of the association between sibship size and height is gradually weakened from the 1840s until the 1940s because of the decline in fertility. The relationship between resource dilution and health therefore seems to be also dependent on the societal and historical context just like the conditional resource dilution hypothesis argues. Until now, only a small number of studies have investigated whether the actual presence of similarly or differently aged siblings was characterized by competition in the way the resource dilution model proposes for infant and child mortality outcomes. Most of these studies investigated, however, contemporary populations and looked at the impact of sibling composition on young girls. One of the first observations was that sibship size was positively associated to infant mortality (Knodel and Hermalin 1984) and that a child’s gender in the sibling set was key for child survival (Bhargava 2003; Choe et al. 1995; Council and Das Gupta 1987; Garg and Morduch 1998; Muhuri and Menken 1997; Pande 2003; Sear et al. 2002). In general, a gendered resource dilution model would imply that gender

1 There are studies who investigate resource dilution on later-life-mortality, such as Barclay and Kolk (2015) and Donrovich,

Puschmann, and Matthijs (2014) and other who study other aspects of health such as (Robinson 2012).

6 preferences are taken into account when allocating parental resources. In most societies this implies that daughters will have fewer resources compared to boys because of son preference. Moreover, the presence of brothers will have the largest effect on the dilution of resources for both boys and girls because the number of sons in a family determines the actual division of resources (Kalmijn and van de Werfhorst 2016). Muhuri and Preston (1991) also estimated that in Matlab (Bangladesh) girls with older sisters faced a large disadvantage and explains a large part of girls’ excess mortality. Later born boys were, however, also discriminated when older brothers are present in the sibling group. More recently Chamarbagwala (2011) found similar results when investigating India. This indicates that next to having a strong son preference, in some contexts parents apparently also seek to balance between the number

• f sons and daughters by selectively discriminating against some children more than others.

Historical studies looking at sibling composition and gender were mostly conducted for Asian societies because in general it was assumed that in historical Europe there were no social or societal encouragements to commit female infanticide or gender bias in the allocation of resources as was the case in many places in Asia (Lynch 2011). However, a major problem is that studies investigating historical Asian populations cannot include young girls when investigating infant mortality because of data limitations: often they are simply not recorded. As a result, the actual position of siblings within the total household remains unobserved (Bengtsson et al. 2004; Dong et al. 2016; Park et al. forthcoming). Other studies attempted to investigate the effects of siblings, but did not ask the same questions about gender composition to Western populations. Kok, Vandezande, and Mandemakers (2011), for example, examined the relationship between household structure – the presence of certain family members – and child mortality when both parents were alive or one or both parents were absent in nineteenth century Netherlands.2 Their main finding was that kin seemed to function the same in nuclear and stem regions within the Netherlands: a 20 percent increase in the share of youth increased and 20 percent increase in the share of elderly decreases infant and child mortality chances. Moreover, they did not find any significant effects for the presence of brother or sisters aged nine or older. These findings are illustrative, but gender differences between boys and girls are excluded just like an explicit focus on sibling effects. Kippen and Walters (2012) on the other hand investigated the influence of sibling composition in diverse age groups and controlled for birth interval and death clustering by using event history analysis. They conclude that the presence of any additional sibling under the age of five increases the probability of dying for children younger than five. In addition, they exclude the sex of the siblings because they found no effect of it on child survival chances. Because of these results they argue that (the presence of) siblings seem to have concrete influences on mortality outcomes, most likely through resource dilution. Yet, only the village of Sart in Belgium is studied and they do not give clear explanations for their findings.

2 Specification makes the model identical to a common logistic regression.

7 A few studies have shown how brothers in the pre-transitional period had different influences

• n male and female child mortality chances compared to sisters (Breschi, Derosas, and Manfredini 2004;

Oris et al. 2004). Derosas (2012) investigates the case of Venice using event history methods and demonstrates that a sibling’s age and gender plays an important and differential role for infant girls and

• boys. They conclude that after parents had reached a certain number and composition of children they

allocated less time and resources to newborns. Different results were found in a study on rural Italy were a gender differential in mortality favoring boys did exist after infancy (Manfredini, Breschi, and Fornasin 2017). Lastly, Fox et al. (2016) study sibling effects on child survival and reproductive success in historical Krummhörn and Quebec by using non-fixed and fixed effects event history models. Their findings demonstrate that having additional siblings had negative effects when the family effect was controlled for. In addition, gender also seemed to play an important role because brothers had the strongest effect for boys, while sisters had the strongest effect for girls. The general consistency and robustness of the sibling effects across different ecological and economic contexts was their most interesting result, but they do not elaborate on what these findings regarding sibling effects on infant and child mortality chances mean in these specific contexts. In sum, while the resource dilution model predicts that every additional child will increase infant and child mortality (H1), , much research has pointed out that the sex of one’s siblings may also have differential effects on young boys’ and girls’ mortality chances. The gendered resource dilution model departs from the assumption that there is a strong son preference. Sons are seen as more valuable than girls because they can inherit the family farm and provide for their parents when they are older. The presence of brothers will therefore automatically result in more resource dilution for both sexes (H2a). Other recent studies have, however, shown that in some contexts parents might try to balance the number and gender composition of their children because there is a clear division of labor between the sexes and both are needed within a well-functioning household. In practice this means that there is no clear son

• preference. If this hypothesis is true for the Netherlands the presence of brothers should be harmful for

boys, while the presence of sisters is harmful for girls (H2b). This study, goes a step further by also using the ideas of the conditional resource dilution model: three regions within the Netherlands are studied to be able to test if the effect of siblings on infant and child mortality is generally the same in each region or that roles and expectations differ between family systems. It does so by building on the idea that family systems are connected with economic conditions, cultural codes and practices, co- residence patterns and family relationships which influence demographic outcomes such as inequality in survival chances. How these issues are connected to each other and might cause different effects of sibling size and composition is explored in the next section.

8

• 3. Historical Context and Family Systems in Nineteenth Century the Netherlands

In general Dutch households were, just like most North-Western European families, relatively small. The parental couple was of central importance and as a result children were expected to set up their own household when they married or when they got children of their own (Hajnal 1965). Yet, household structure and organization also varied considerably across regions and over time within the Netherlands, thereby causing regional differences in family systems that defined parent-child and sibling

• relationships. Earlier studies already pointed at these regional household patterns and defined three

broad regions based on co-residence: the nuclear north-west (Noord-Holland, Zuid-Holland, Groningen, Drenthe and Friesland), the intermediate south-west (Utrecht, Zeeland and Noord-Brabant), and the stem south-east (Limburg, Gelderland and Overijssel).3 The main differences – which are connected to each

• ther – between these three broad regions are suggested to result from economic conditions such as

economic agricultural developments and inheritance practices (Kok and Mandemakers 2008, 2010; Verduin 1985). Firstly, economic agriculture developments were different across the Netherlands because of the differences in soil types, the nature of the type of agriculture and agricultural traditions. Agriculture was on the one hand relatively small-scale and profitable in the commercialized north-western part of the Netherlands. As a result, family labor was not as important for farmers because help could be hired when it was needed. On the other hand, in the other parts of the country agriculture remained a self- sustaining family enterprise for a much longer period which caused farmers to rely on the labor of family members much more (Van Zanden 1985; Knippenberg & Pater 1990: 97-106). These differences also had an influence on the age and type of work children (from the age of eight) had to do within or outside the household to help make ends meet. Mostly this meant that sons and daughters were working together with their parents in the family business or on the farm. Even most schoolchildren performed all kinds

• f jobs before, in between, and after school hours, on the weekends and during their holidays. The kind
• f work children had to do, however, different in most regions and was gendered in nature (Gates 2005;

De Regt 2004; Schenkeveld 2008). The type of work was different because in the pasture areas in North Holland, South Holland, Utrecht and Friesland most farms were capital-intensive on not so labor-intensive. The result was that the amount of work that could be done by children was relatively limited. Children mostly had to bring the animals to the field and help harvesting the hay. The milking of cows was done by both boys and girls, but the skills of cheese making were mostly transferred from mother to daughter. On the sandy soils in the south-east of the Netherlands such as Noord-Brabant, Limburg, Gelderland and Overijssel the opposite was the case and farms were much more labor-intensive. Many children had to help in one way or another and it was also the main reason fewer children attended school in these areas. Next to

3 It should be noted that even in the defined stem regions the majority of the households were still nuclear.

9 working on the parental farm, children also had to watch their younger siblings and do household chores so that their mothers could work on the family farm or get a small job outside the household. The latter was mostly done by girls. Daughters helped their mothers with the household chores, minded younger brothers and sisters, knitted and sewed, while older sons often had a job that earned them a wage or they worked on the family farm. Moreover, in general girls had less time for playing than their brothers because girls began serious necessary work on an earlier age and because of the idea that their work was never truly done (Gates 2005; Schenkeveld 2008). Lastly, by contributing a substantial amount of money to the family budget children could become irreplaceable as co-providers and were able to claim some more resources or freedom in return for the money they brought in (Klep 2011; De Regt 2004) Secondly, while with the introduction of the French Civil Code in 1811 partible inheritance was legalized everywhere in the Netherlands, reality was different because farmers transmitted their property according to regional norms and customs of property transfer. Especially for the eastern part of the Netherlands (Achterhoek, Twente, Salland) it is clear that often the father choose one designated heir (often the youngest son) who stayed on the farm and became the head of the household. Parents decided when the heir would officially take over the farm and helped on the farm as long as they could. Because

• f the unequal distribution of the inheritance, unmarried siblings had a life-long right to stay on the farm.

In addition, this pattern was also found among labourers, although less frequently. In other words, there was a strong connection between the stem family, impartibility and care contracts in the east of the

• Netherlands. In other parts of the Netherlands, however, inheritance practices had different
• consequences. In the southern parts the inheritance was equally divided, but only after both parents had
• died. In most instances children could therefore only start an own farm with their savings and through
• marriage. Yet, marriage ages and permanent celibacy were high because of this later fragmentation of
• property. Many unmarried siblings remained together to work on the parental farm as a result. In central

parts, care contracts were mentioned in combination with partibility, whereas in the southwest the geographic location of many farms simply prevented their division. This forced impartibility may have stimulated co-residence as well (Bras and Van Tilburg 2007; Kok and Mandemakers 2008, 2010). The above mentioned differences in economic conditions influenced the presence (household structure) and power (household relations) of kin within the household which defined the three earlier mentioned co-residence patterns. Moreover, it is also argued that these regional differences are not caused solely by economic conditions. Cultural codes and practices were entwined with economic conditions and also played a role through historical family systems which originate from earlier times, but are still influential through generations of socialization (Bras and Van Tilburg 2007). The work of Todd (1985) offers a good starting point to incorporate the concept of family systems with the conditional resource dilution model because it focuses both on generational and intergenerational

• relationships. He argues that family systems vary along two axes indicating liberty versus authority on

the on hand and equality versus inequality on the other hand. Relations between father and son

10 (intergenerational relations) determine the degree of liberty, while the bond between brothers (generational relations) creates the degree of equality. With this framework four family systems are created and distinguished: the absolute nuclear family (liberal parent-child and inegalitarian sibling relations), the egalitarian nuclear family (liberal parent-child and egalitarian sibling relations), the authoritarian family (authoritarian parent-child and inegalitarian sibling relations) and the communitarian family (authoritarian child-parent and egalitarian sibling relations).4 However, Todd (1985) recognizes that in the case of absolute nuclear family equality between children is not an important ideal and that these households are indifferent to equality between brothers and to male solidarity (Todd 1985: 28). This means that the absolute nuclear family has in taken equality between the sexes further than the egalitarian nuclear family because the principle of solidarity between brothers implies masculine solidarity and reinforces inequality between the sexes. In Table 1 the typology of Todd (1985) – with taking the cultural codes and practices and economic conditions into account – is applied to the three mentioned regions in the Netherlands. Because in the nuclear area children are not needed as much to work inside or outside the household and inheritance is mainly divided on the basis on wills, children may be viewed more as consumers than producers which leads to resource dilution for every additional child. Furthermore, indifference to gender indicates one the one hand that the dilution of resources takes place whenever an additional child is born no matter its gender, while on the other hand an additional brother could also

• nly impact the amount of resources that are available for boys and an additional sister only impact the

amount of resources that are available for girls. If the latter is true and the former is not it might cause more sex specific resource dilution because parents only need a certain number of girls and boys (H3a). Compared to the nuclear area, the stem area should show most contrasting results as children were much more needed to do some work within the household or on the family farm. Children generally also stayed longer within the household because of the inheritance practices which resulted in different (less individual) parent-child and sibling relations. Parents are more likely to favor the heir over other siblings and cause sibling relations to be unequal, causing especially resource dilution for non-heir children (H3c). Lastly, in the intermediate family region economic conditions are mixed and it seems difficult to really infer any consequences for resource dilution. Yet, if indeed all material resources are equally divided between all siblings, which would mostly be the case for the present older brothers, any additional brother will cause resource dilution and increase mortality chances for boys and girls (H3b).

4 Todd bases much of his work on an adopted typology of Le Play, but argues it was incomplete because he did not include the

liberal and egalitarian type of family. Moreover, Todd also argues that Le Play did not take into account if marriage was exogamous or endogamous. In his work he therefore further extended the typology to 7 types of family systems which do include these issues (Todd 1985: 7-32).

11 Table 1: Summary Characteristics of the three regions within the Netherlands

• 4. Data, Measurements, and Methods

For this research the Historical Sample of the Netherlands (HSN), which compiled life course data as completely as possible from a sample of the birth registers of nineteenth and early twentieth-century Dutch society, is used. The sample of the HSN for this paper consists of all the research persons who were born between the period 1863 and 1905 from the Life Courses Release 2010.01 Data Set. For this time period we have access to exact dates of birth and death during the first five years of life, and the sample covers the whole country. Important to note is that at this moment only one person from each household is selected as a research person. This means that there are no observations which are tied

• together. A further selection was made by only including those research persons who did not experience

the death of a parent. In other words, both parents had to be alive during the first five years of the research person’s life. This is done because sibling roles may differ when a parent is missing and this study focuses on sibling effects in ‘complete’ families. A second selection was made by excluding research persons who were twins, because the proportional hazard assumption in the Cox proportional- hazard models was not met when this variables was included. This should not lead to major shortcomings because in general twins have higher mortality chances and the number of research persons who are twins is low. Most longitudinal life course data from the HSN was extracted from Dutch population registers. These registers were created by Royal Decree to start recording information in population registers from 1850 onwards. The main reason for the government to implement this kind of registration was because

Region Agricultural condition Inheritance practices Family System (classification Todd) Parent-Child Relations Sibling Relations Sibling Effects on Child Mortality Nuclear family region Small-scale, profitable and capital-intensive Partible with the use of wills Absolute nuclear Liberal Indifferent towards gender Any additional sibling (boy or girl) will cause resource dilution and increase mortality chances for

• children. This may also be specific for
• nly same-sex siblings (H3a)

Intermediate family region Mix between capital and labor- intensive Mix between partible and impartible Egalitarian nuclear Liberal/Authoritari an Equal between brothers Any additional brother will cause resource dilution and increase mortality chances for boys and girls (H3b) Stem family region Self-sustaining family enterprise and labor- intensive Mostly one heir or equally divided after parents died Authoritarian Authoritarian Unequal between brothers Having no older brother reduces resource dilution and decreases mortality chances. For any other additional siblings besides the heir resource dilution is limited because of equal division of resources (H3c).

12 it would become possible to track changes in population between censuses. It therefore recorded the head of the household and all other household members such as a wife, children, other relatives, boarders and so on. The relationship to the head of the household, gender, marital status, occupation, religion and place and date of birth were recorded for every individual (Knotter & Meijer, 1995). The important advantage of this data is that it records exact dates when household composition changes. For example, when new household members arrived or were born after the registration had started they were added to the already recorded list of individuals, and household members who died or moved were removed with reference to date of death or place and data of migration (Mandemakers 2006). The ability to follow individuals on a day-to-day basis is key because what matters for resource dilution is not the biological birth placement, but the number of the siblings each child has during the period of parental investment. In a pre-transitional society the biological birth order and actual birth order within a household can differ a lot because of high mortality, marriage and migration. Neglecting this changing size of the household has led to shortcomings in previous studies, but by using reliable longitudinal datasets from the HSN this problem is solved. To analyze all data, methods appropriate for the statistical analysis of quantitative life course data are used. The univariate analysis is done by using Kaplan-Meier survival curves, while multivariate event history Cox proportional-hazard models can analyze mortality chances by taking time and control variables into account (Cleves 2008; Broström 2012). Observation time begins with the birth of a child and ends when it dies (failure event) before his or her fifth birthday. All remaining children (which are research persons) are censored on their fifth birthday. Hazard ratios calculated using the Cox proportional-hazard model are presented in table 5 and 6. The hazard ratio is the influence of the given category on the mortality hazard relative to the reference category, controlling for all other variables. Probability values of less than 0.10 are assumed to represent statistical significance. Diagnostics were run to check if all assumptions were met. In most cases this was not a problem, and when the assumptions were not fully met other models were run which confirmed the same qualitative conclusions. The main variables of interest are the number of siblings on child mortality. The number of siblings a research person is defined as having the same father or mother. These variables are controlled for by including birth interval5, mother’s age, whether the previous infant dies6, father’s occupation7, religion8, if the location was urban or rural9 and period in the final model. Separate models are used for neonatal (0-1 month), post-neonatal (1-12 months) and young child mortality (12-60 months) because the interaction

5 Several different categories are used in recent research, but in the end the qualitative conclusions stay the same. For the models

in this study the classification of Engelen and Wolf (2011) was used with the adjustment of adding a category for firstborns.

6 Also categories for death clustering - from Kippen and Walters (2012) and Nault, Desjardins, and Légaré (1990) - were used

but these did not make a difference for the effects of brothers and sisters.

7 First the standard HISCO classification was used – see Van Leeuwen, Maas, and Miles (2004) - and then categories of elite

and middleclass were combined to form the category of professionals because of low numbers in the category elite.

8 See Kok (2017) for more information on the classification of liberal and orthodox protestants. 9 Urban communities are defined as places with over 10000 inhabitants and with less than two

and a half percent of the population employed in the agricultural sector. See Kooij (1985) for more information.

13

• f endogenous and exogenous mortality determinants changes throughout childhood and covariates are

likely to have different effects as a results during each of these age categories. Important to note is that the number of co-resident siblings is treated as a time-varying covariate. This accommodates the fact that siblings could move in and out of the household while the index child was growing up. In addition, the number of siblings are divided in age categories compared to the research person: younger and older siblings. These groups are based on the assumption that the age of the present siblings may be of importance. Gender is also included by explicitly looking at the number

• f brothers and sisters in each age category. Moreover, the analysis was done by using three different

models to see if there were any mediating or confounding factors. In the first model the number of brother and sisters is included together with indicators of infant health: birth intervals, mother’s age and previous sibling dies. In the second model also the occupation of the father and religion are added to control for economic conditions and cultural norms. In the final model indicators for the kind of family system, whether a place was urban or not and the period are added to control for the time and place in which the research persons lived. The summary statistics of all variables – measured at the moment when the research persons were born – can be found in Table 2. Table 2: Description of the variables when the research person is born (1863-1905) N % Dies % Dies Gender Male 10970 51.29 2883 26.281 Female 10415 48.70 2395 22.99 Mother Age <20 160 0.74 40 25.00 20-35 14320 67.02 3348 23.38 >=35 6884 32.22 1877 27.26 Social economic status Unskilled 6761 31.61 1792 26.50 Skilled 6197 28.97 1625 26.22 Professional 3020 14.12 659 21.82 Farmers 2288 10.69 502 21.94 Unknown 3119 14.58 700 14.58 Previes infant dies No 19542 91.38 4647 23.78 Yes 1843 8.61 631 34.23 Birth Interval <16 2931 13.70 982 33.50 16-24 3841 17.96 998 25.98 >24 6370 29.78 1336 20.97 First born 8243 38.54 1962 23.80 Region

14 Intermediate 5392 25.21 1389 25.76 Nuclear 11420 53.40 2848 24.93 Stem 4573 21.38 1041 22.76 Religion Liberal protestants 9443 44.15 2179 23.07 Orthodox protestants 3509 16.40 924 26.33 Catholics 7326 34.25 1914 26.12 Other 1107 5.17 261 23.57 Period <1885 9289 43.43 2758 29.69 >1885 12096 56.56 2520 20.83 # Sisters 9685 45.28 2319 23.94 1 5658 26.45 1440 25.45 2 3244 15.17 833 25.67 3 1680 7.85 426 25.35 4+ 1118 5.22 260 23.25 # Brothers 9762 45.64 2321 23.77 1 5406 25.27 1346 24.89 2 3284 15.35 886 26.97 3 1711 8.00 431 25.19 4+ 1222 5.71 294 24.05 Urban No 13768 64.38 3272 23.76 Yes 7617 35.61 2006 26.33 Average % dies 24.681 5.1 Results: Univariate analyses Previous studies have shown that mortality chances differ between place and time in the studied regions (Engelen 2009; Van Poppel 2011). Therefore, to gain an understanding in the differences of the mortality levels in the regions under study within the Netherlands are calculated first (Table 3). For the whole of the Netherlands about 16 percent of the boys and 13 percent of the girls die before age 1. This percentages increases somewhat for children who survive until age five: 26 percent of the boys, and 23 percent of the girls. In all regions boys have higher mortality chances compared to girls, which confirms earlier studies that indicate that boys are more vulnerable at young ages because of biological disadvantages (Bouman, Jan Heineman, and Faas 2005; Mage and Donner 2004; Mage and Maria Donner 2014; Waldron 1983). The mortality chances are lowest for infants (0-12 months) in the stem family region, while in the intermediate region infants have the highest mortality chances. For children (0-60 months) the same observation can be made. This is in line with earlier observations of the mortality trend in the Netherlands where in the nineteenth century mortality levels are higher in the western areas.

15 This higher mortality pattern only starts to move from the west to the east of the Netherlands at the end

• f the nineteenth century.

To get insight in the mortality chances of children with diverse sets of siblings the infant and child mortality chances (0-60 months) of boys and girls are calculated by using Kaplan-Meier curves for the number of brothers (Table 4a) and number of sisters (Table 4b) for each region and for the Netherlands as a whole. It is clear that boys with an older brother have higher mortality chances compared to boys with no brother, but mortality does not always increase for every additional brother. An exception is the stem family region where the differences are not that large, except for boys with two brothers. Still, for the Netherlands as a whole the differences of mortality chances of boys between the number of brothers are significant. For girls the number of brothers does not seem to have a clear harmful or beneficial effect, while the opposite is true for the number of sisters. Having one or more sisters compared to none increases mortality chances for girls in the nuclear and intermediate family. Moreover, the differences in mortality chances for girls between the number of sisters in the nuclear and intermediate region are statistically significant. For boys the number of sisters does not have a clear effect on child mortality chances. The above reported results suggest that if the mechanisms of resource dilution are at work, they work through gender composition. In general, for boys the number of brothers and for girls the number

• f sisters seems to increase mortality chances. Only in the stem family region these results are more
• ambiguous. To investigate this issue further the mortality chances by gender and region are calculated

for four different sibship compositions: no siblings are present, only brothers are present, only sisters are present, and both brothers and sisters are present (Table 4c). Surprisingly, the previous results are confirmed as mortality chances in all regions are higher for boys in families where there are only brothers and for girls where there are only sisters. In addition, for the Netherlands as a whole, for boys in the nuclear region and girls in the intermediate region the differences in mortality chances are statistically significant between the defined sibling sets. The lowest mortality chances are found for children who are born in a household without any siblings or where there are only siblings of the opposite sex. In sum, in these explorative analyses there seems to be a clear gender component in the effect sibling size and composition in general have: the number of same-sex siblings has a clear negative effect. In other words, children who are born in a household in which siblings are of the opposite sex are better

• ff. This suggests that parents might consciously or unconsciously be aiming for gender balance, which

could be explained by the need for a few children of each sex because of gender roles within the

• household. As already mentioned, girls had to do more household chores in general and could contribute

to specific tasks like making cheese, while boys had more options outside the household. Still, these results also seem to indicate that resource dilution could play an important role because having no siblings also resulted in lower mortality chances. Still, there might be confounding variables for which

16 there is no control in these univariate analyses. Moreover, the number and composition of the sibling set is taken when the research person is born and not followed over time. To be able to look into more detail and take changing sibling composition into account, the next section investigates the same sibling effects in a multivariate model by using event history Cox proportional-hazard models and time-varying variables for the number of siblings. Table 3: Mortality Chances of infant and children in the Netherlands as a whole and the selected areas divided by gender and age category of the research person at birth (1863-1910). Table 4a: Mortality Chances of children (0-60 months) in the Netherlands as a whole and the selected areas divided by the number of brothers and gender of the research person at birth (1863-1910).

• Signif. code:* p < 0.1

Table 4b: Mortality Chances of children (0-60 months) in the Netherlands as a whole and the selected areas by the number of sisters and gender of the research person at birth (1863-1910).

• Signif. code:* p < 0.1

Table 4c: Mortality Chances of children (0-60 months) in the Netherlands as a whole and the selected areas by sibling set composition and gender of the research person at birth (1863-1910). Months The Netherlands Nuclear family region Intermediate family region Stem family region Boys Girls Boys Girls Boys Girls Boys Girls 0-1 month 0.023 0.018 0.022 0.017 0.025 0.020 0.025 0.020 1-12 months 0.164 0.134 0.166 0.136 0.177 0.152 0.140 0.107 1-5 years 0.100 0.097 0.101 0.097 0.100 0.087 0.102 0.110 0-5 years 0.264 0.231 0.267 0.233 0.277 0.239 0.242 0.217 Total brothers The Netherlands Nuclear family region Intermediate family region Stem family region Boys* Girls Boys Girls Boys Girls Boys Girls 0.250 0.229 0.253 0.229 0.255 0.234 0.236 0.223 1 0.271 0.228 0.269 0.235 0.302 0.240 0.233 0.192 2 0.289 0.252 0.296 0.242 0.286 0.268 0.280 0.266 3 0.269 0.236 0.274 0.250 0.274 0.243 0.245 0.186 4+ 0.278 0.203 0.298 0.210 0.271 0.195 0.238 0.200 Total sisters The Netherlands Nuclear family region Intermediate family region Stem family region Boys Girls Boys Girls* Boys Girls* Boys Girls 0.263 0.218 0.269 0.222 0.259 0.213 0.251 0.214 1 0.268 0.243 0.272 0.232 0.278 0.279 0.246 0.222 2 0.271 0.244 0.264 0.267 0.314 0.228 0.225 0.205 3 0.259 0.249 0.277 0.270 0.285 0.231 0.184 0.230 4+ 0.246 0.220 0.219 0.191 0.263 0.255 0.293 0.229

17

• Signif. code:* p < 0.1

5.2 Results: Cox proportional-hazard models The influence of the number brothers and sisters on male and female neonatal (0-1 month), post-neonatal (1-12 months) and child (13-60 months) mortality chances in the Netherlands as a whole is explored by using three different models (Table 5). A first general observation is that the direction of the effect of all control variables on mortality chances are as expected and in line with earlier research and the results

• f Kok et al. (2011). In the first model the number of brothers and sisters is included together with

endogenous indicators of infant health: birth intervals, mother’s age and previous sibling dies. Being first born or having a shorter birth interval than 16 months was associated with higher mortality chances, while having a longer birth interval than 24 months led to lower mortality chances compared to having a birth interval between 16 and 24 months. Having a mother below 20 or over 35 years old also increased mortality chances. The effect of birth interval and mother’s age suggests that maternal depletion played a major role, but that it affected boys more compared to girls. This is as expected because of the different

• rganic and biological resistance of males and females to illness, with girls having a stronger immune

response to infection during the first year of life (Conde-Agudelo et al. 2012; Engelen and Wolf 2011; Kozuki and Walker 2013; Winkvist, Rasmussen, and Habicht 1992). When a previous sibling died as an infant this had only an increased effect on mortality chances of both infants girls and boys, although this result is not statistical significant for neonatal mortality for boys. In sum, variables that act as an indicator for maternal depletion or overall health of the household suggest that these determinants played a major role in determining infant and child mortality chances In the second model also the occupation of the father and religion are added to control for economic conditions and religious cultural norms and practices. Compared to the social-economic group of children whose fathers were unskilled workers, children of fathers in the group professionals and farmers had decreased post-neonatal and child mortality chances. Only for neonatal mortality there was no clear effect, except for children of farmers having higher neonatal mortality changes, which was only slightly statistically significant for boys. This suggests that available material resources, including nutritional intake because of the amount, diversity and quality of food, only decreased mortality chances after the first month of life. Although the results of children’s religious affiliation does not provide a clear picture, it does show that post-neonatal and child mortality chances are higher for Catholic girls compared to liberal protestant girls. Orthodox protestant children also have higher post-neonatal mortality chances Sibling set The Netherlands Nuclear family region Intermediate family region Stem family region Boys* Girls* Boys* Girls Boys Girls* Boys Girls No siblings 0.246 0.214 0.252 0.221 0.235 0.196 0.242 0.213 Only brothers 0.292 0.226 0.301 0.225 0.288 0.236 0.271 0.216 Only sisters 0.256 0.255 0.256 0.246 0.281 0.283 0.226 0.244 Mix 0.270 0.237 0.270 0.243 0.290 0.245 0.238 0.208

18 compared to liberal protestant children. Although other research finds similar effects, there is not one consistent explanation regarding specific child care practices for each religion. They only exception are lower motility chances of Jews, which are not included as a separate category (Van den Boomen and Ekamper 2015; Poppel 1992). In the final model indicators for the kind of family system, whether a place was urban or not and the period are added to control for the time and place in which the research persons lived. In contrast to what Kok et. al (2011) found there seems to be a statistically significant mortality differences between the nuclear, intermediate and stem family regions. The effects are, however, reversed for different age categories: in the stem family region neonatal mortality chances are lower for both girls and boys, while the child mortality chances are higher for girls. Moreover, for girls child mortality chances are also higher in nuclear family regions. The reversed direction and mediating effect of gender may be the reasons why this result was not found in earlier studies where child mortality was not divided in different age categories. Children living in cities had higher mortality chances than children living in rural areas after the first months of life. Lastly, a child in the period 1863-1885 experienced higher mortality chances compared to children in the period 1885-1910), which results from the decline in mortality starting in the nineteenth century (Van Poppel 2011). A second important observation can be made regarding the variable of interest: the interpretation of the effect of the number of brothers and sisters stays the same in all three models with different

• covariates. This indicates that the included factors were not mediating or confounding the effect of the

number of children present as for example might have been expected when households with a certain social-economic status or religion had more children. The neonatal mortality chances for boys were lower when more older sisters or brothers were present, while for girls the same can be observed for the number of older sisters. Yet, it is unlikely that altruism could already play a role at this age because mortality in the first month of life is mostly the result from endogenous determinants. The presence of brothers and sisters demonstrates, in my opinion, that the research person is born in a healthier household because there are already other siblings alive. As is also well known, infant death tended to cluster in some families, while infant survival clustered in others (Gupta 1990; Vandezande 2012). Still, it is striking that when having same-sex siblings differences in mortality chances are statistically significant. For post-neonatal mortality there are no statistical significant results, expect higher mortality chances for girls having older sisters. In general these findings for infant mortality are not surprising because support or rivalry of siblings would still only play a small role because infants were still being breastfed most of the time.

19 Table 5: Cox proportional-hazard models of child mortality divided by gender in the Netherlands (1863-1910)

20 The results for child mortality chances are the most interesting because the number of older brothers had a negative effect on mortality chances, but only for boys. The number of sisters had no significant influence on child mortality for boys and also for girls there seems no clear impact of the number of siblings of either sex. The same observation can be made when the analysis is done for the different birth cohorts (Table 6). While for girls the effect of the number of brothers and sisters does not seem to have a clear trend, an increasing influence over time is visible for the influence of the number of brothers for boys’ mortality chances. In the cohorts of 1860s and 1870s the higher mortality chances of boys with

• ne or more brothers is not statistically significant, but in the cohorts 1880s to 1900s it becomes

statistically significant and also increases in strength. It indicates that when mortality declined and fertility also slowly declined the number of brothers had a bigger impact on mortality chances. Table 6: Cox proportional-hazard model (including all control variables) of child mortality divided by cohorts in the Netherlands (1863-1910)

• Signif. codes: '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1

Above described results indicate again that resource dilution might play a role, but that this role is

• gendered. The results, however, do differ a bit from the univariate analysis in the sense that after adding

control variables not both sexes experience higher mortality chances when there are more same-sex siblings, but that when one or more brothers are present mortality chances are only statistically significant higher for boys. To further investigate the conditional research model the influence of the number of brothers and sisters on neonatal, post-neonatal and child mortality is examined in more detail by differentiating between nuclear, intermediate and stem family regions (Table 7). Nevertheless, because the number of research persons is lower in each model and all control variables are the same as those of the final model in Table 5, strong statistically significant results can hardly be expected. Still, similar results for neonatal and post-neonatal mortality chances are found (not shown here). For child mortality the results show both similarity and difference between the three regions. In the nuclear family region the same statistically significant results can be found as for the whole of the Netherlands: the presence of older brothers causes higher child mortality chances. In addition, boys with one or more younger brothers also have higher child mortality chances. The same observation can be made for the intermediate and stem family region when looking at the direction of the effect of the

21 number of brothers, but only in the intermediate family region the number of older brothers causes also significantly higher mortality chances. A difference in child mortality chances of boys between the regions can be found in the effect of the number of younger sisters for boys in the stem family region which are lower. This could point at a beneficial effect of being a child with a lower birth order. Yet, the results for child mortality chances for girls suggest that they experienced the biggest difference in child mortality chances by region when having more brothers or sisters. In the nuclear family region

• lder sisters significantly increase child mortality chances for girls, while in other regions this effect is

the opposite (although not statistically significant). This seems to suggest that in nuclear areas there is more resource dilution or competition when same-sex siblings are present as also girls are effected. Table 7: Cox-proportional-hazard model (including all control variables) of child mortality divided by nuclear, intermediate and stem family area in the Netherlands (1863-1910)

• Signif. codes: '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1
• 6. Preliminary Discussion and Conclusion

Research on the relation between child mortality and household structure and organization is moving from the general assumption of the resource dilution model to more nuanced views such as the conditional or gendered resource dilution model. The aim of this study was to use these concepts to examine the impact of the size and composition of the sibling set on infant and child mortality chances. By doing so, the position and gender of individuals within household hierarchies were examined in three different regions and throughout time in the Netherlands. These regions were chosen because the conditional research model suggests that even within a small country like the Netherlands different economic conditions and inheritance practices influenced the way in which household resources are divided among children. In turn, these differences in family systems may also influence sibling relations to the extent that they shape the most extreme form of inequality: survival chances. Because the research persons are scattered over a large are and a broad time period not all results

• f this study are easy to understand. It is clear that next to socio-cultural factors also biological factors

shaped child mortality chances, especially for infants. The most important finding, however, was that Sex Male Male Male Female Female Female Model 3 3 3 3 3 3 Age in months 13-60 13-60 13-60 13-60 13-60 13-60 Population Nuclear Intermediate Stem Nuclear Intermediate Stem Variables # Older Sisters (TVC) 0.935 0.990 0.925 1.087+ 0.928 0.949 # Younger Sisters (TVC) 1.100 1.161 0.563* 1.124 0.707 1.114 # Older Brothers (TVC) 1.114* 1.144* 1.115 1.051 0.983 0.926 # Younger Brothers (TVC) 1.243+ 0.998 0.687 1.089 0.887 0.926

22 for the Netherlands as a whole it seemed that only boys were negatively influenced by the presence of

• lder brothers regarding their child mortality chances. It is in line with results found in the earlier

discussed studies that argue that parents might try to balance the number and gender composition of their children because of the absence of son preference and specific roles within and outside the household assigned to each gender. A difference with these studies is that for the Netherlands as a whole resource dilution only occurred for boys with same-sex siblings which supports the suggestion from a recent study that Dutch couples were actually balancing the composition of the sibling set from an early stage in contrast to other countries and preferred girls over boys (Reher et al. 2017). They base their argument on the observation that primarily the couples with no surviving girls exhibited increased birth intensities, suggesting a stronger tendency for preferring girls in the surviving sibling set. To explain this preference these authors point at the idea that from the seventeenth and eighteenth century onwards the Dutch Republic provided more economic and social space for women than in other European societies, leading to a relatively more emancipated position (Moor and Van Zanden 2010). A remaining question, however, is in what way the formulated hypotheses can explain the differences and similarities between the three regions within the Netherlands. Only in the nuclear family region the number of same-sex siblings had a negative impact on child mortality chances for both boys and girls. This is in line with the formulated hypothesis and follows the logic of a region in which parents are indifferent to gender and where most farms and industry are capital-intensive. In other words, both sons and daughters offered similar resources for the household while it only needed a limited number of children of each gender. Still, the observation that both older and younger brothers caused higher mortality chances for boys suggests that there was much more competition for resources among sons and that age did not matter as much. In the intermediate family area something similar could be the case, but only the number of brothers causes higher mortality chances among boys. This is in line with the expectation that brothers are treated equally and therefore have to share resources in the egalitarian nuclear family. In contrast with the situation in the nuclear family area, in the stem family area the number of brothers and sisters did not have any statistically significant harmful effects on survival

• chances. This could result from the fact that these households were more collective orientated and could

still use the labor of their children. An additional daughter could especially always be welcome to offer a helping hand on in the household. In sum, this study has firstly demonstrated the importance of including the number of siblings in child mortality research. It can enrich our perspective on the determinants of well-being in general and the processes of sibling competition and support specific. Most studies only include birth order or sibling composition at birth, but this study demonstrated that it is important to include the changing household composition over time. It is crucial to use the number of siblings as a time-varying variable to be able to take the actual number and of the siblings in the household into account as the difference between the results of the univariate and cox proportional-hazard models have shown. Secondly, this

23 study showed that sibling size and composition through resource dilution – even when both parents are alive – may have an important impact on children’s survival chances. This study has, however, also tried to go one step further to be able to demonstrate that the exact influence of sibling size and composition depends on the specific economic conditions, cultural codes and practices, and family systems as the conditional resource dilution model suggests. Regional differences, even within a small country like the Netherlands, are even confirmed which indicates that reality is more complex than the ‘simple’ resource dilution model suggests. Future studies should therefore take these considerations into account when studying sibling effect on mortality and other demographic outcomes. In addition, more instructive findings of mediating effects of family systems on infant and child mortality chances through sibling effects may be found when stronger variations of the absolute nuclear, egalitarian nuclear, authoritarian and communitarian family regions are investigated. My PhD-project will therefore also compare these results with several communities in Taiwan were two variations of the communitarian family can be found.

24 References

Bailey, By R. O. Y. E., Timothy J. Hatton, and Kris Inwood. n.d. “Health, Height, and the Household at the Turn of the Twentieth Century

†.” 1:35–53. Barclay, Kieron and Martin Kolk. 2015. “Birth Order and Mortality: A Population-Based Cohort Study.” Demography 52(2):613–39. Retrieved March 31, 2015 (http://www.ncbi.nlm.nih.gov/pubmed/25777302). Beekink, Erik and Jan Kok. 2016. “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.” The History of the Family 5398(August):1–18. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2016.1212722). Bengtsson, Tommy, Cameron Campbell, and James Z. Lee. 2004a. Life under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900. Bengtsson, Tommy, Cameron Campbell, and James Z. Lee. 2004b. “Life under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900.” MIT Press Books 1. Bhargava, Alok. 2003. “Family Planning, Gender Differences and Infant Mortality: Evidence from Uttar Pradesh, India.” Journal of Econometrics 112(1):225–40. Black, Sandra E., Paul J. Devereux, and Kjell G. Salvanes. 2005. “The More the Merrier? The Effect of Family Size and Birth Order on Children’s Education.” The Quarterly Journal of Economics 669–700. van den Boomen, Nynke and Peter Ekamper. 2015. “Denied Their ‘natural Nourishment’: Religion, Causes of Death and Infant Mortality in the Netherlands, 1875–1899.” The History of the Family (April 2015):1–29. Retrieved April 13, 2015 (http://www.tandfonline.com/doi/full/10.1080/1081602X.2015.1022199). Booth, Alison L. and Hiau Joo Kee. 2008. Birth Order Matters: The Effect of Family Size and Birth Order on Educational Attainment. Retrieved March 31, 2015 (http://link.springer.com/10.1007/s00148-007-0181-4). Bouman, Annechien, Maas Jan Heineman, and Marijke M. Faas. 2005. “Sex Hormones and the Immune Response in Humans.” Human Reproduction Update 11(4):411–23. Bras, H. and T. Van Tilburg. 2007. “Kinship and Social Networks: A Regional Analysis of Sibling Relations in Twentieth-Century Netherlands.” Journal of Family History 32(3):296–322. Bras, Hilde, Jan Kok, and Kees Mandemakers. 2010. “Sibship Size and Status Attainment across Contexts: Evidence from the Netherlands, 1840-1925.” Demographic Research 23:73–104. Breschi, M., R. Derosas, and M. Manfredini. 2004. “Mortality and Environment in Three Emilian, Tuscan, and Venetian Communities, 1800-1883.” Pp. 209–52 in Life Under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900, edited by T. Bengtsson, C. Campbell, and J. Z. Lee. Cambridge: MIT Press. Bronfenbrenner, Urie. 1979. The Ecology of Human Development. Harvard university press. Bronfenbrenner, Urie. 1986. “Ecology of the Family as a Context for Human Development: Research Perspectives.” Developmental Psychology 22(6):723. Broström, Göran. 2012. Event History Analysis with R. CRC Press. Carter, Scott, Lala Carr Steelman, Brian Powell, Regina Werum, and Scott Carter. 2002. “Reconsidering the Effects of Sibling Configuration: Recent Advances and Challenges.” Annual Review of Sociology 28(1):243–69. Retrieved February 18, 2015 (http://www.annualreviews.org/doi/abs/10.1146/annurev.soc.28.111301.093304). Chamarbagwala, Rubiana. 2011. “Sibling Composition and Selective Gender-Based Survival Bias.” Journal of Population Economics 24(3):935–55. Retrieved August 6, 2014 (http://link.springer.com/10.1007/s00148-010-0314-z). Choe, Minja Kim, H. Hao, F. Wang, Hao Hongsheng, and Wang Feng. 1995. “Effects of Gender, Birth Order, and Other Correlates on Childhood Mortality in China.” Social Biology 42(1–2):50–64. Chu, C. Y.Cyrus, Y. Xie, and R.-r. Yu. 2007. “Effects of Sibship Structure Revisited: Evidence from Intrafamily Resource Transfer in Taiwan.” Sociology of Education 80(2):91–113. Cleves, Mario. 2008. An Introduction to Survival Analysis Using Stata. Stata Press. Conde-Agudelo, Agustin, Anyeli Rosas-Bermúdez, Fabio Castano, and Maureen H. Norton. 2012. “Effects of Birth Spacing on Maternal, Perinatal, Infant, and Child Health: A Systematic Review of Causal Mechanisms.” Studies In Family Planning 43(2):93–114. Conley, Dalton. 2000. “Sibship Sex Composition: Effects on Educational Attainment.” Social Science Research 29(3):441–57. Retrieved (http://www.sciencedirect.com/science/article/pii/S0049089X00906781%5Cnhttp://www.sciencedirect.com/science?_ob=MiamiImag eURL&_cid=272353&_user=2459739&_pii=S0049089X00906781&_check=y&_origin=article&_zone=toolbar&_coverDate=30- Sep-2000&view=c&originContentF).

25

Council, Population and Monica Das Gupta. 1987. “Selective Discrimination against Female Children in Rural Punjab, India.” Populations and Development Review 13(1):77–100. Derosas, Renzo. 2012. Suspicious Deaths: Household Composition, Infant Neglect, and Child Care in Nineteenth-Century Venice. Retrieved (http://www.cairn.info/revue-annales-de-demographie-historique-2012-1-page-95.htm). Derosas, Renzo and Michel Oris. 2002. When Dad Died. Individuals and Families Coping with Family Stress in Past Societies. Bern: Peter Lang. Dong, Hao, Matteo Manfredini, Satomi Kurosu, Wenshan Yang, and James Z. Lee. 2016. “Kin and Birth Order Effects on Male Child Mortality: Three East Asian Populations, 1716–1945.” Evolution and Human Behavior 1–9. Retrieved (http://linkinghub.elsevier.com/retrieve/pii/S1090513816300678). Donrovich, Robyn, Paul Puschmann, and Koen Matthijs. 2014. “Rivalry, Solidarity, and Longevity among Siblings.” Demographic Research 31(November):1167–98. Retrieved March 31, 2015 (http://www.demographic-research.org/volumes/vol31/38/). Downey, Douglas B. 1995. “When Bigger Is Not Better: Family Size, Parental Resources, and Children’s Educational Preformance.” American Sociological Review 60(5):746–61. Downey, Douglas B. 2001. “Number of Siblings and Intellectual Development: The Resource Dilution Explanation.” American Psychologist 56(6–7):497. Dribe, Martin, Cameron Campbell, and Jan Van Bavel. 2012. “Social Mobility and Demographic Behaviour: Long Term Perspectives.” Demographic Research 26:173–90. Retrieved February 25, 2015 (http://www.demographic-research.org/volumes/vol26/8/). Engelen, Th L. M. 2009. Van 2 Naar 16 Miljoen Mensen. Demografie van Nederland, 1800-Nu. Amsterdam: Boom. Engelen, Th L. M. and A. P. Wolf. 2011. “Maternal Depletion and Infant Mortality.” Fox, Jonathan, Kai P. Willführ, Alain Gagnon, Lisa Dillon, and Eckart Voland. 2016. “The Consequences of Sibling Rivalry on Survival and Reproductive Success across Different Ecological Contexts : A Comparison of the Historical Krummhörn and Quebec Populations.” The History of the Family 5398(0):1–60. Retrieved (http://dx.doi.org/10.1080/1081602X.2016.1193551). Garg, A. and J. Morduch. 1998. “Sibling Rivalry and the Gender Gap: Evidence from Child Health Outcomes in Ghana.” Journal of Population Economics 11(4):471–93. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/12294784). Gates, Hill. 2005. “Girls’ Work in China and North-Western Europe: Of Guniang and Meisjes.” Pp. 319–41 in Marriage and the family in

• Eurasia. Perspectives on the Hajnal hypothesis, edited by T. Engelen and A. P. Wolf. Amsterdam: Aksant.

Gibbs, Benjamin G., Joseph Workman, and Douglas B. Downey. 2016. “The (Conditional) Resource Dilution Model: State- and Community-Level Modifications.” Demography 53(3):723–48. Retrieved (http://dx.doi.org/10.1007/s13524-016-0471-0). Goody, Jack. 1996. “Comparing Family Systems in Europe and Asia: Are There Different Sets of Rules?” Population and Development Review 22(1):1-2-. Gupta, Monica Das. 1990. “Death Clustering, Mothers’ Education and the Determinants of Child Mortality in Rural Punjab, India.” Population Studies 44(3):1–1. Das Gupta, Monica. 1997. “Kinship Systems and Demographic Regimes.” Anthropological Demography: Toward a New Synthesis, Ed. David I. Kertzer and Tom Fricke (Chicago, IL: The University of Chicago Press, 1997) 36–52. Das Gupta, Monica. 1999. “LIFEBOAT VERSUS CORPORATE ETHIC: Social and Demographic Implications of Stem and Joint Families.” Social Science & Medicine 49(2):173–84. Hajnal, John. 1982. “Two Kinds of Preindustiral Household Formation System.” Population and Development Review 8(3):449–94. Hatton, Timothy J. 2016. “Stature and Sibship : Historical Evidence Stature and Sibship : Historical Evidence.” 5398(June):0–21. Hilevych, Yuliya. 2016. “Strong Families and Declining Fertility: A Comparative Study of Family Relations and Reproductive Careers in Soviet Ukraine.” Wageningen University. Hrdy, Sarah Blaffer. 2011. Mothers and Others: The Evolutionary Origins of Mutual Understanding. Harvard University Press. Kalmijn, Matthijs and Herman G. van de Werfhorst. 2016. “Sibship Size and Gendered Resource Dilution in Different Societal Contexts.” Plos One 11(8):e0160953. Retrieved (http://dx.plos.org/10.1371/journal.pone.0160953). Keyser, Lieselot De and Ronan Van Rossem. 2016. “The History of the Family Birth Order and Height Development in Early Childhood : Empirical Evidence for Twenty-First- Century Flanders.” 5398(August). Kippen, Rebecca and Sarah Walters. 2012. “Is Sibling Rivalry Fatal? Siblings and Mortality Clustering.” Journal of Interdisciplinary History 42(4):571–91. Klep, Paul M. M. 2005. “An Adult Life before Marriage: Children and the Hajnal Hypothesis.” Pp. 241–70 in Marriage and Family Living,

26

edited by T. Engelen and A. P. Wolf. Amsterdam: Aksant. Klep, Paul M. M. 2011. Gezinssolidariteit En Rotten Kids. Schaarste, Seks En Het Vierde Gebod Op Het Platteland in Nederland in de Twintigste Eeuw. Nijmegen: University Nijmegen. Knodel, J. and a I. Hermalin. 1984. “Effects of Birth Rank, Maternal Age, Birth Interval, and Sibship Size on Infant and Child Mortality: Evidence from 18th and 19th Century Reproductive Histories.” American Journal of Public Health 74(10):1098–1106. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1651865&tool=pmcentrez&rendertype=abstract). Kok, Jan. 2017. “Church Affiliation and Life Course Transitions in The Netherlands, 1850-1970.” Historical Social Research 42(2):59–91. Kok, Jan A. N. and Kees Mandemakers. 2010. “A Life-Course Approach to Co-Residence in the Netherlands, 1850–1940.” Continuity and Change 25(2):285–312. Retrieved August 6, 2014 (http://www.journals.cambridge.org/abstract_S0268416010000160). Kok, Jan and Kees Mandemakers. 2008. “"Je Zou Maar Last van Mij Hebben’’. Verwanten in Het Nederlandse Huishouden, 1860-1940.” Tijdschrift Voor Sociale En Economische Geschiedenis 6(4):139–65. Kok, Jan, Matthijs Vandezande, and Kees Mandemakers. 2011. “Household Structure, Resource Allocation and Child Well-Being: A Comparison Across Family Systems.” Tijdschrift Voor Sociale En Economische Geschiedenis 8(4):76–101. Kooij, P. 1985. “Stad En Platteland.” Pp. 93–115 in De Nederlandse samenleving sinds 1815. Wording en samenhang, edited by F. L. van

• Holthoon. Assen: Van Gorcum.

Kozuki, Naoko and Neff Walker. 2013. “Exploring the Association between Short/long Preceding Birth Intervals and Child Mortality: Using Reference Birth Interval Children of the Same Mother as Comparison.” BMC Public Health 13 Suppl 3(Suppl 3):S6. Retrieved August 6, 2014 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3847658&tool=pmcentrez&rendertype=abstract). Van Leeuwen, Marco H. D., Ineke Maas, and Andrew Miles. 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):186–97. Li, Hongbin, Junsen Zhang, and Y. I. Zhu. 2008. “THE QUANTITY-QUALITY TRADE-OFF OF CHILDREN IN A DEVELOPING COUNTRY : IDENTIFICATION USING CHINESE TWINS *.” 45(1):223–43. Lu, Yao, Donald J. Treiman, and Yao Lu Donald. 2014. “The Effect of Sibship Size on Educational Attainment in China : Period Variations.” 73(5):813–34. Lynch, Katherine A. 2011. “Why Weren’t (Many) European Women ‘missing’?” The History of the Family 16(3):250–66. Mage, D. T. and E. M. Donner. 2004. “The Fifty Percent Male Excess of Infant Respiratory Mortality.” Acta Paediatrica (Oslo, Norway : 1992) 93(9):1210–15. Mage, David T. and E. Maria Donner. 2014. “Is Excess Male Infant Mortality from Sudden Infant Death Syndrome and Other Respiratory Diseases X-Linked?” Acta Paediatrica, International Journal of Paediatrics 103(2):188–93. Mandemakers, Kees. 2006. “Building Life Course Datasets from Population Registers by the Historical Sample of the Netherlands (HSN).” History and Computing 14(1–2):87–107. Manfredini, Matteo, Marco Breschi, and Alessio Fornasin. 2017. “MORTALITY DIFFERENTIALS BY GENDER IN THE FIRST YEARS OF LIFE: THE EFFECT OF HOUSEHOLD STRUCTURE IN CASALGUIDI, 1819-1859.” The Journal of the Economic & Business History Society XXXV(1):291–314. Retrieved (http://myaccess.library.utoronto.ca/login?url=http://search.proquest.com/docview/302915958?accountid=14771%5Cnhttp://bf4dv7z n3u.search.serialssolutions.com/?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF- 8&rfr_id=info:sid/ProQuest+Dissertations+%26+Theses+G). Marteleto, Letícia J. and Laetícia R. de Souza. 2012. “The Changing Impact of Family Size on Adolescents’ Schooling: Assessing the Exogenous Variation in Fertility Using Twins in Brazil.” Demography 49(4):1453–77. Mason, Karen Oppenheim. 2001. “Gender and Family Systems in the Fertility Transition.” Population and Development Review 27:160–76. Mazzoni, Stanislao, Marco Breschi, Matteo Manfredini, Lucia Pozzi, and Gabriele Ruiu. 2016. “The Relationship between Family Characteristics and Height in Sardinia at the Turn of the Twentieth Century.” The History of the Family 5398(October):1–19. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2016.1230509). Mönkediek, Bastian. 2016. “Family Systems and Fertility. Fertility Behaviour in Europe from a Network Perspective.” Wageningen University. Moor, Tine de and Jan Luiten Van Zanden. 2010. “‘ Every Woman Counts ’: A Gender-Analysis of Numeracy in the Low Countries during the Early Modern Period.” The Journal of Interdisciplinary History 41(2):179–208. Muhuri, Pradip K. and Jane Menken. 1997. “Adverse Effects of Next Birth, Gender, and Family Composition on Child Survival in Rural Bangladesh.” Population Studies 51(3):279–94. Muhuri, Pradip K. and Samuel H. Preston. 1991. “Effects of Family Composition on Mortality Differentials by Sex among Children in

27

Matlab, Bangladesh.” The Population and Development Review 415–34. Myrskylä, Mikko, Karri Silventoinen, Aline Jelenkovic, Per Tynelius, and Finn Rasmussen. 2013. “The Association between Height and Birth Order: Evidence from 652,518 Swedish Men.” Journal of Epidemiology and Community Health 67(7):571–77. Retrieved March 31, 2015 (http://www.ncbi.nlm.nih.gov/pubmed/23645856). Nault, F., B. Desjardins, and J. Légaré. 1990. “Effects of Reproductive Behaviour on Infant Mortality of French-Canadians during the Seventeenth and Eighteenth Centuries.” Population Studies 44(2):273–85. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/11622323). Öberg, Stefan. 2015. “Sibship Size and Height Before, During, and after the Fertility Decline.” Demographic Research 32(January):29–74. Retrieved February 15, 2015 (http://www.demographic-research.org/volumes/vol32/2/). Öberg, Stefan. 2017. “Too Many Is Not Enough: Studying How Children Are Affected by Their Number of Siblings and Resource Dilution in Families.” The History of the Family 22(2–3):157–74. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2017.1302890). Oris, Michel, Renzo Derosas, M. Breschi., Marco Breschi, and others. 2004. “Infant and Child Mortality.” Pp. 359–98 in Life Under Pressure: Mortality and Living Standards in Europe and Asia, 1700-1900, edited by T. Bengtsson, C. Campbell, and J. Z. Lee. Cambridge: MIT Press. Pande, Rohini. 2003. “Nutrition and Immunization in Rural India : The Role of Siblings *.” Population (English Edition) 40(3):395–418. Poppel, Frans van. 1992. “Religion and Health: Catholicism and Regional Mortality Differences in Nineteenth-Century Netherlands.” Social History of Medicine. Van Poppel, Frans. 2000. “Children in One-Parent Families: Survival as an Indicator of the Role of the Parents.” Journal of Family History 25(3):269–90. Van Poppel, Frans. 2011. “Trends in Mortality and the Evolution of the Cause-of-Death Pattern in the Netherlands: 1850-2000.” Pp. 17–43 in Death at the opposite ends of the eurasian continent. Mortality trends in Taiwan and the Netherlands, 1850-1945, edited by T. Engelen, J. R. Shepherd, and W.-S. Yang. Amsterdam: Aksant. Poppel, Frans Van and Ruben Van Gaalen. 2008. “The Presence of Parents and Childhood Survival: The Passage of Social Time and Differences by Social Class.” Kinship and Demographic Behavior in the Past 7(2):105–34. Retrieved (http://www.springerlink.com/index/g88170020ukg0337.pdf). Poulain, Michel, Dany Chambre, Anne Herm, and Gianni Pes. 2017. “Anthropometric Traits at Military Medical Examinations Associated with Demographic Family Characteristics in Sardinia at the Turn of Twentieth Century.” The History of the Family 22(2–3):310–32. Retrieved (http://dx.doi.org/10.1080/1081602X.2016.1193819). Powell, Brian, Regina Werum, and Lala Carr Steelman. 2004. “Macro Causes, Micro Effects: Linking Public Policy, Family Structure, and Educational Outcomes.” After the Bell-Family Background, Public Policy, and Educational Success 111–44. Ramon-Muñoz, Ramon and Josep-Maria Ramon-Muñoz. 2016. “Sibship Size and the Biological Standard of Living in Industrial Catalonia, c.1860–c.1920: A Case Study.” The History of the Family 5398(September):1–31. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2016.1224727). De Regt, Ali. 2004. “Children in the 20th-Century Family Economy: From Co-Providers to Consumers.” History of the Family 9(4):371–84. Reher, David Sven. 1998. “Family Ties in Western Europe: Persistent Contrasts.” Population and Development Review 24(2):203–34. Reher, David Sven, Glenn Sandström, Alberto Sanz-Gimeno, and Frans W. A. van Poppel. 2017. “Agency in Fertility Decisions in Western Europe During the Demographic Transition: A Comparative Perspective.” Demography 54(1):3–22. Retrieved (http://dx.doi.org/10.1007/s13524-016-0536-0). Roberts, Evan and John Robert Warren. 2016. “Family Structure and Childhood Anthropometry in Saint Paul, Minnesota in 1918.” The History of the Family 5398(October):1–33. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2016.1224729). Robinson, Christina. 2012. “Younger Siblings Can Be Good for Your Health: An Examination of Spillover Benefits from the Supplemental Nutrition Program for Women, Infants, and Children (WIC).” Journal of Family and Economic Issues 34(2):172–84. Retrieved March 31, 2015 (http://link.springer.com/10.1007/s10834-012-9325-0). Schenkeveld, Willemien. 2008. “Het Werk van Kinderen in de Nederlandse Landbouw 1800-1913.” Tijdschrift Voor Sociale En Economische Geschiedenis 5(2):28–54. Sear, R. and R. Mace. 2008. “Who Keeps Children Alive? A Review of the Effects of Kin on Child Survival.” Evolution and Human Behavior 29(1):1–18. Retrieved August 6, 2014 (http://linkinghub.elsevier.com/retrieve/pii/S1090513807001055). Sear, Rebecca and David Coall. 2011. “How Much Does Family Matter? Cooperative Breeding and the Demographic Transition.” Population and Development Review 37:81–112. Sear, Rebecca, Fiona Steele, Ian A. McGregor, and Ruth Mace. 2002. “The Effects of Kin on Child Mortality in Rural Gambia.” Demography 39(1):43–63. Retrieved

28

(http://link.springer.com/article/10.1353/dem.2002.0010%5CnC:%5CUsers%5Clgaydosh%5CAppData%5CRoaming%5CZotero%5 CZotero%5CProfiles%5Cep00jnl9.default%5Czotero%5Cstorage%5CDMNI5DGF%5Cdem.2002.html). Shavit, Yossi, Jennifer L. Pierce, and L. Pierce. 1991. “Sibship Size and Educational Attainment in Nuclear and Extended Families: Arabs and Jews in Isreal.” American Sociological Review 56(3):321–30. Skinner, G.William. 1997. “Family Systems and Demographic Processes.” Pp. 53–95 in Anthropological demography: Toward a new

• synthesis. Chicago: University of Chicago Press.

Stradford, Laura, Frans van Poppel, and L. H. Lumey. 2016. “Can Resource Dilution Explain Differences in Height by Birth Order and Family Size? A Study of 389,287 Male Recruits in Twentieth-Century Netherlands.” The History of the Family 5398(October):1–22. Retrieved (https://www.tandfonline.com/doi/full/10.1080/1081602X.2016.1230510). Therborn, Göran. 2004. Between Sex and Power: Family in the World 1900-2000. Routledge. Todd, Emmanuel. 1985. The Explanation of Ideology: Family Structures and Social Systems. Blackwell Oxford. Vandezande, Mattijs. 2012. “Born to Die? Intergenerational Patterns of Infant Mortality. A Research Agenda.” Status: Published. Verduin, J. A. 1985. “Het Gezin in Demografisch Perspectief.” Pp. 69–110 in Gezinsgeschiedenis. Vier eeuwen gezin in Nederland, edited by

• G. A. Kooy. Assen and Maastricht.

Waldron, Ingrid. 1983. “The Role of Genetic and Biological Factors in Sex Differences in Mortality.” in Sex Differentials in Mortality: Trends, Determinants and Concequences, edited by A. Lopez and L. Ruzicka. Canberra: Canberra Australia Australian National University Department of Demography. Winkvist, a, K. M. Rasmussen, and J. P. Habicht. 1992. “A New Definition of Maternal Depletion Syndrome.” American Journal of Public Health 82(5):691–94. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1694126&tool=pmcentrez&rendertype=abstract). Wolf, Arthur P. 2005. “Europe and China: Two Kinds of Patriarchy.” Pp. 215–40 in Marriage and the family in Eurasia. Perspectives on the Hajnal hypothesis, edited by T. Engelen and A. P. Wolf. Amsterdam: Aksant. Yu, Wei-hsin, Kuo-hsien Su, and Chi-Tsun Chiu. 2012. “Sibship Characteristics and Transition to First Marriage in Taiwan: Explaining Gender Asymmetries.” Population Research and Policy Review 31(4):609–36. Retrieved March 31, 2015 (http://link.springer.com/10.1007/s11113-012-9236-7).