The Patriarchy Index: A New Measure of Gender and Generational Inequalities in the Past

Abstract

This article presents a new measure of family-driven age- and gender-related inequalities. This composite measure, which we call the Patriarchy Index, combines a range of variables related to familial behavior that reflect varying degrees of sex- and age-related social inequality across different family settings. We demonstrate the comparative advantages of the index by showing how 266 historical populations from the Atlantic coast of Europe to Moscow scored on the patriarchy scale. We then compare the index with contemporary measures of gender discrimination, and find a strong correlation between historical and current inequality patterns. Finally, we explore how variation in patriarchy levels across Europe is related to the socioeconomic and institutional characteristics of the regional populations, and to variation across these regions in their degree of demographic centrality and environmental conditions. The results confirm previous findings that family organization is a crucial generator of social inequality, and point to the importance of considering the historical context when analyzing the current global contours of inequality.

Keywords

patriarchy household and family systems living arrangements historical GIS gender inequality power relations developmental inequalities historical census microdata

Introduction

Inequality is one of the most discussed issues in contemporary social sciences, and in national and global politics (Milanovic, 2005). Over the past decade, the study of inequality has advanced considerably. Large quantities of data have been collected on a (nearly) global scale, and increasingly sophisticated analyses of these data have been conducted. The aim of these analyses has been to identify the different dimensions of “inequality,” the processes that led to divergence and convergence in these trends, and the consequences of existing disparities (Therborn, 2006). Economists in particular have devoted a great deal of energy to conducting global studies of the contemporary distributional dynamics. Economic historians and demographers have also contributed to this discussion by generating findings that have improved our understanding of historical inequalities in the spatial distribution of wealth, income, and well-being around the globe (e.g., Klüsener et al., 2014; Milanovic, Lindert, & Williamson, 2011; van Zanden, Baten, Foldvari, & van Leeuwen, 2014; van Zanden, Baten, Mira d’Ercole, et al., 2014).

Gender inequality has always been a crucial element in these debates. Interest in the issue of gender equality has been fueled by the recognition that women play important roles in a wide range of development outcomes (World Bank, 2011). While we have extensive scientific evidence on broad movements aimed at achieving gender equality in many parts of the world, whether we are able to monitor the impact of these efforts efficiently depends on our ability to measure forms of sex-related inequality across societies. Over the past three decades, specific measures have been developed that capture different aspects of gender inequality in outcomes, and that focus on the institutions that perpetuate gender disparities (for a review, see Klasen 2006; Malhotra, Sidney, & Boender, 2002; recently Carmichael, Dilli, & Rijpma, 2014; Dilli, Rijpma, & Carmichael, 2015).

Although the body of literature on gender inequality is large, most of the existing studies on this issue have at least two main drawbacks. First, there is a striking absence of long-term perspective in many quantitative gender inequality studies. None of the composite gender indices used in the developmental literature pre-date the 1990s, and while recent attempts to provide greater historical depth via the Historical Gender Equality Index (HGEI) undeniably further the current measurement spectrum, they represent only a moderate step forward in terms of providing a long-term perspective, as the gender measures they are based on go back only as far as the 1950s (see Carmichael et al., 2014; Dilli et al., 2015). When we move further back in time, the available indicators tend to narrow quite dramatically, and are nonexistent for larger social groups and geographical population clusters in the more distant past (Drwenski, 2015). This lack of historical data is a potentially serious problem for scholars of contemporary trends in gender inequality, because variation in levels of gender inequality may have historical roots, and the processes through which women have gained greater equality in terms of rights and socioeconomic standing have unfolded over a long period of time (Dorius & Firebaugh, 2010).

Second, while various authors have stressed the multidimensionality of gender inequality (e.g., Mason, 1986; Schlegel, 1972; Whyte, 1978; Young, Fort, & Danner, 1994), most of the analyses have tended to investigate gender discrimination separately from other associated forms of discrimination. However, it has been suggested that gender inequality is inextricably intertwined with other systems of inequality (Coltrane & Adams, 2000; cf. also Young et al., 1994), and especially with discrimination by age (seniority), i.e., the institutionalized superiority of older family members relative to younger family members. According to Therborn (2004), age discrimination and gender inequality are the two “basic intrinsic dimensions” of patriarchy (pp. 13-14). These two forms of expropriation are dialectically related, and often act to reinforce each other in fostering a complex hierarchy of authority patterns based on both age and gender (see Dyson & Moore, 1983; Joseph, 1996; also Halpern, Kaser, & Wagner, 1996).

In this article, we make a fourfold contribution to this ongoing discussion. First, we propose the use of a new inequality measure: namely, the Patriarchy Index (PI), which encompasses gender and its related discrimination dimension (i.e., seniority; see Gruber & Szołtysek, 2016). Second, we apply this new measure to examine regional patterns, thereby providing geographical richness and temporal depth to existing accounts of gender and generational inequalities in the European past. Third, we demonstrate that variation in the PI across Europe is highly correlated with spatial variation in contemporary measures of gender inequality, which suggests that variation in gender equality across Europe is subject to path dependencies. A fourth distinctive contribution of our article is that we attempt to explore the relevance of specific socioeconomic, institutional, and locational characteristics for explanations of historical variation in patriarchy across Europe.

The text is organized as follows. First, we present our data and explain how they were used for the construction of the PI. Next, we illustrate how the PI is applied to data for 266 regional populations of historic Europe, located from the Atlantic coast to Moscow. In two subsequent sections, we check for correlations between our measure and other gender inequality measures, and then present a spatially sensitive regression analysis of the relationship between patriarchy levels and broad variations in socioeconomic, institutional, and locational characteristics across Europe. We conclude by highlighting the implications of our findings for research on historical inequality patterns and comparative development.

Data

Because historical measures of inequality are difficult to develop, any attempts to study this issue on a larger scale are contingent upon the availability of relevant data (Johnston, 1985). In constructing a composite historical measure of age and sex discrimination, we relied on census and census-like microdata. We chose these data because of their broad availability across historic Europe. Thanks to the Mosaic Project and the North Atlantic Population Project (NAPP), such data are publicly available in the form of machine-readable, harmonized microdata samples that are relatively easy to process (Ruggles, Roberts, Sarkar, & Sobek, 2011; Szołtysek, 2015a; Szołtysek & Gruber, 2016; Table 1 and Figure 1 show the distribution of regions covered by Mosaic and NAPP across Europe).¹

Table 1.

Data Used for Analysis.

Census	Regions	N (=pop.)
Mosaic data
Albania, 1918 census	8 rural regions, 6 cities	140,611
Austria-Hungary, 1869 census	9 rural regions from Hungary, Romania, and Slovakia	31,406
Austria-Hungary, 1910 census	3 rural regions and 1 city from Austria	20,036
Belgium, 1814 census	1 rural region from western Flanders	13,666
Bulgaria, 1877-1947 household registers	1 rural region and 1 city from the Rhodope area	8,373
Dubrovnik, 1674 status animarum	1 rural region from Dalmatia	1,880
Denmark, 1803 census	9 rural regions and 2 urban regions from Schleswig and Holstein	107,861
France, 1846 census	3 rural regions	16,967
France, 1831-1901 census	1 rural region from southwestern France	5,109
France, 1846-1856 census	1 city from southwestern France	5,669
German Customs Union, 1846 census	10 rural regions and 4 urban regions	36,760
German Customs Union, 1858 census	1 rural region from the east	3,468
German Customs Union, 1861 census	1 rural region from the southwest	6,541
German Customs Union, 1867 census	4 rural regions and 1 city in Mecklenburg-Schwerin	66,938
Germany, 1900 census	1 city	55,705
Mecklenburg-Schwerin, 1819 census	3 rural regions and 1 city	37,332
Münster, around 1700 status animarum	3 rural regions in northwestern Germany	23,010
Münster, 1749 status animarum	3 rural regions in northwestern Germany	34,169
Netherlands, census 1810-1811	2 rural regions and 3 cities in the south	40,037
Poland-Lithuania, 1768-1804 listings	12 rural regions	155,818
Moldavia, 1781-1879 status animarum	2 rural regions	5,291
Wallachia, 1838 census	4 rural regions	21,546
Russia, 1795 revision lists	1 rural region in Ukraine	8,050
Russia, 1814 private enumeration	1 region in central Russia	2,955
Russia, 1847 enumeration	2 rural regions in Lithuania and Belarus	19,917
Russia, 1897 census	1 rural region around Moscow	11,559
Serbia, 1863 census	1 rural region and 1 city	9,746
Serbia, 1884 census	1 rural region	9,434
Spain, 1880-1890 local census	1 rural and 2 urban regions in Catalonia	23,997
Ottoman Empire, 1885 census	Istanbul	3,408
Ottoman Empire, 1907 census	Istanbul	4,946
Mosaic data overall	115 regions (89 rural and 26 urban)	932,205
NAPP data
Denmark, 1787 census (100%)	21 regions	838,623
Iceland, 1703 census (100%)	1 region	51,003
Norway, 1801 census (100%)	19 regions	878,073
Sweden, 1880 census (100%)	24 regions	4,624,825
The United Kingdom, 1881 census
England (10% sample)	76 regions	2,926,374
Wales (100% sample)	13 regions	1,573,065
Scotland (100% sample)	32 regions	2,783,354
Islands (100% sample)	3 regions	139,614
NAPP data overall	151 regions	14,252,150

Source. Mosaic/NAPP data. For primary sources of the Mosaic and NAPP data: see Online Appendix 2.

Note. NAPP = North Atlantic Population Project.

Figure 1.

Spatial distribution of Mosaic and NAPP data by major territorial groupings.

The Mosaic Project (Szołtysek & Gruber, 2016) currently encompasses 115 regional populations of continental Europe captured through various kinds of historical census and census-like materials other than full-count national censuses (e.g., local fragments of censuses, church lists of parishioners, tax lists, local estate inventories). It contains data going back to 1700, or even earlier.² To cover Great Britain and Scandinavia, we also decided to draw upon historical national census public-use microdata from NAPP (see Ruggles et al., 2011).³ The Mosaic and the NAPP microdata samples are very similar in terms of structure, organization, and the types of information they provide. These samples describe the characteristics of individuals in a given settlement or area grouped into households (coresident domestic groups), and provide information on the relationships between coresident individuals. All the demographic variables stored in these two datasets are harmonized across space and time using common international standards, which allows us to generate historical localized gender and generational indicators across multiple locations.

As we situate our approach at the meso level of comparative analysis, our units of analysis are “regions.” The regions in the NAPP data are the administrative units that were used in the respective census and that were considered by NAPP. The Mosaic data are organized by separate locations, which in most cases also represent separate administrative units. However, as the Mosaic data for a given region are often not complete, and as we lack information on the exact administrative boundaries of many of them, we had to use more flexible approaches (see Szołtysek & Gruber, 2016). As a rule of thumb, we ensured that each Mosaic region had at least 2,000 inhabitants, and that urban and rural settlements were separated. In a few cases, enumeration data from the same unit (usually urban) collected at different time periods are treated as independent regions. Overall, our analysis covers 266 regional populations (see Table 1).

We grouped these regions into seven larger territorial clusters designed to capture the range of institutional and socioeconomic characteristics across Europe. The NAPP data were used in the Scandinavia and the Great Britain clusters. The Mosaic data were divided into the following clusters: Germany (German-dominated areas other than the Habsburg territories); West (areas west and southwest of Germany); Habsburg, East (central-eastern and eastern Europe, including the former Polish-Lithuanian Commonwealth and Russia); and Balkans (areas south and/or east of Croatia and Hungary). In the regression analysis, we further subdivided some of these regions when introducing regional dummies. Our motivation for this decision was a desire to account for within-region variation in the PI levels, which remained unexplained in the models with all the covariates other than the regional dummies. The Great Britain cluster was subdivided into England, Scotland, and Wales; the East cluster into Central-East (Poland) and East (locations further east); and the Balkans cluster into Albania and Southeast (Figure 1).

The 266 regional populations cover large parts of Europe, and run across many—though not all⁴—important fault lines in the European geography of demographic regimes (Hajnal, 1982; Szołtysek, 2015a). Furthermore, our dataset covers a large share of the variation across Europe in terms of geographical features, populations, cultures, and socioeconomic geography: that is, plains, mountains, and coastal areas; the free and the unfree peasantries; a variety of ethnicities and religions; and a range of regional patterns of economic growth in the early modern and modern eras. Of the 266 regional datasets, a slight majority (59%) represents populations after 1850, while 41% cover populations before 1850, and 16% populations before 1800. The collection includes information on both rural and urban sites, although rural societies clearly predominate.⁵

Patriarchy and Its Composite Measure

In line with a number of recent theorists, we see patriarchy not as having a single form or site, but as encompassing a much wider realm (cf. Joseph, 1996; Kandiyoti, 1988). According to Therborn (2004), patriarchy has two basic intrinsic dimensions: “the rule of the father and the rule of the husband, in that order” (pp. 13-14). Thus, patriarchy encompasses both stratification of social attainment by sex and the domination of men over each other based on the seniority principle (Joseph, 1996). Halpern, for example, showed that the multifaceted nature of the Balkan patriarchy was historically anchored in the interlocking combination of the rule of the father, the eldest man in the family, and the husband (Halpern et al., 1996).⁶

Based on these considerations, we conceptualized “patriarchal” elements as clustering in the four “domains” that we believe capture the four major dimensions of the phenomenon under consideration: the domination of men over women, the domination of the older generation over the younger generation, the extent of patrilocality, and the preference for sons. Table 2 provides a list of the components we considered. The table also shows how we defined and measured these components, and it indicates the expected direction of their relationship with societal patriarchy levels (+/−) (for a comprehensive discussion of all components and age standardization, see Gruber & Szołtysek, 2016).

Table 2.

Components of the Patriarchy Index.

Domain	Component	Abbreviation	Definition/measurement	Relationship with patriarchy	Specification
Male domination	Proportion of female household heads	Female heads	The proportion of all female household heads (20+ years) among all adult heads of family households	Negative	Age-standardized
	Proportion of young brides	Young brides	The proportion of ever-married women in the age group 15-19 years	Positive
	Proportion of wives who are older than their husbands	Older wives	The proportion of all wives who are older than their husbands among all couples for whom the ages of both spouses are known	Negative	Age-standardized
	Proportion of young women living as nonkin	Female nonkin	The proportion of women aged 20-34 years who live as nonkin, usually as lodgers or servants	Negative	Age-standardized
Generational domination	Proportion of elderly men coresiding with a younger household head	Younger household head	The proportion of men aged 65+ years living in a household headed by a male household head of a younger generation	Negative	Only family households; the elderly men must be relatives of the household head
	Proportion of neolocal residence among young men	Neolocal	The proportion of male household heads living without any relatives except spouses/children among ever-married men aged 20-29 years	Negative	Only family households; age-standardized
	Proportion of elderly people living with lateral relatives	Lateral	The proportion of people aged 65+ years living with at least one lateral relative in the household	Positive	Only family households
Patrilocality	Proportion of elderly people living with married daughters	Married daughter	The proportion of people aged 65+ years living with at least one married daughter in the same household among those elderly people who live with at least one married child in the same household	Negative	Only family households
Son preference	Proportion of boys among the last child	Boy as last child	The proportion of boys among the last children (if the last child is one of a set of siblings of both sexes, he or she will be excluded from the analysis)	Positive	Only children (aged 10-14 years) of household heads; family households
	Sex ratio of youngest age group	Sex ratio	The sex ratio (boys to 100 girls) in the youngest age group (0-4 years old)	Positive	Only family households

We chose these components because we believe they capture the most essential aspects of particular domains, given the data constraints. Most of the component variables directly account for various forms of gender and generational biases at the household level. Other variables, like patrilocality, proxy behavioral patterns that could not be derived directly from our data (in this case, inheritance practices). For the most part, we chose to use individual-level age-specific measures instead of household-level variables, because the former tend to minimize the undesired influence of variation in demographic conditions on indicators of family structure (Szołtysek, 2015b). For example, instead of using the incidence of three-generation households for the domain generational domination, we decided to account for the generational patterns of headship, the age-specific patterns of household formation, and the residential patterns of the aged. Accordingly, we chose not to consider the proportion of the elderly living with a married son (another common demographic measure), because without the inclusion of information on headship, it is a poor measure of the level of patriarchal behavior in the domestic group.

From our component variables, we derive the PI as a single composite measure, following the strategy detailed elsewhere (Gruber & Szołtysek, 2016). The PI characterizes the situations of women, the aged, and young people according to the extent to which they had obtained socially valued resources (such as a desirable position or status); albeit without measuring the positions of these groups relative to certain normative standards or reference categories. The index values thus represent absolute, not relative, measures of gender and age inequality (see Johnston, 1985; Young et al., 1994). Table 3 presents a summary of the descriptive statistics for all the variables considered for the computation of the index.

Table 3.

Descriptive Statistics for the Components of the Patriarchy Index for 266 Regional Populations of Europe (Mosaic/NAPP combined).

Component	M	SD	Minimum	Maximum
Female heads	0.13	0.06	0.01	0.31
Young brides	0.06	0.11	0.00	0.66
Older wives	0.22	0.08	0.01	0.37
Female nonkin	0.21	0.12	0.00	0.57
Younger household head	0.13	0.11	0.00	0.68
Neolocal	0.60	0.23	0.03	0.97
Lateral	0.14	0.12	0.00	0.73
Married daughter	0.35	0.20	0.00	0.80
Boy as last child	0.50	0.05	0.34	0.81
Sex ratio	101.90	7.80	81.80	137.30

Source. As in Table 1; own calculations.

Note. NAPP = North Atlantic Population Project; M = mean; SD = standard deviation.

The index’s domains were shown to be positively correlated with each other at a significant level (although none of these correlations were exceptionally high; Gruber & Szołtysek, 2016). As we can see in Figure 2, similar results were found for the relationship between gender and generational domination; variables that we assume are intertwined. This point needs to be emphasized in the context of Todd’s (1987) assertion that high female status and strong parental authority over children were often present simultaneously in historic populations, and led to increased investments in human capital. However, the empirical evidence we provide in Figure 2 does not show the presence of a high degree of female agency combined with a high degree of parental power. Overall, the within-index relationships we found are reassuring, as they both validate the application of this variable as a measurement of patriarchy, and justify our claim that it is important to explore gender and seniority biases in conjunction with each other.

Figure 2.

Scatterplot of the Male Domination Index and the Older Generation Domination Index (Mosaic/NAPP combined).

Deriving gender and generational biases from domestic coresidence data—like the data used in the Mosaic/NAPP format—has certain theoretical merits. The household not only played an essential role in the functioning of preindustrial economies and societies (Szołtysek, 2015a); it represented the most basic arena in which kinship bonds were formed, socialization occurred, and values were transmitted. Those values were concerned with issues of power and equality, justice and gender relations, age hierarchy, and the relationship between the individual and the authorities (Kok, 2014). Because family and household organization patterns affected the status of women, the level of investment in human capital, and the persistence of specific cultural norms and values, it is perhaps not surprising that there is a strong relationship between prevalent family structures and development levels of regions (e.g., Alesina & Giuliano, 2014; Carmichael, Dilli, & van Zanden, 2016). It is thus clear that the household is a particularly meaningful site for measuring gender equity and discrimination (e.g., Folbre, 1986; Malhotra et al., 2002; also Narayan, 2006; Carmichael & Rijpma, 2017).

Spatial Distribution of the PI

We present the distribution of the PI across space in two ways. Figure 3 charts the data geographically, while Figure 4 shows the complete scale of index points arranged according to macroregional membership and time period.

Figure 3.

The spatial distribution of the Patriarchy Index (Mosaic/NAPP combined).

Figure 4.

Regional values of the Patriarchy Index by time period and macrogeographical membership (Mosaic/NAPP combined).

The observed PI values range from 8 to 35 points. In the context of the data we used, we found that while all the regional populations had at least some patriarchal features, as defined above, none of the regional populations could be characterized as fully patriarchal (maximum PI: 40 points). At the most general level, the ranking of the regions is broadly consistent with previous findings from the historical demographic and sociological literature, and seems to confirm the well-known east-west pattern (Hajnal, 1982; Therborn, 2004). Western Europe was shown to be much less patriarchal than eastern and southeastern Europe. If we look at the map (Figure 3), we can see that patriarchal features become increasingly prevalent as we move east and south of the Danube after it passes Vienna; and east of the Bug River, a tributary of the Vistula river, where Polish and Ukrainian ethnicities converge; and then farther into the territories of European Russia.

This generalization is, however, subject to some qualifications. While it is indeed the case that the areas around the North Sea Basin had relatively low patriarchy levels, similarly low levels were also found in parts of Germany and the areas of Scandinavia near the Baltic Sea. Especially in the cities in today’s eastern Germany, the levels of patriarchy appear to have been low. Indeed, patriarchy levels were low in regions spread across a vast area of Europe, ranging from Iceland and Great Britain; through northern France, the Low Countries, and parts of Germany and Scandinavia; into Poland and Austria. Equally interesting is the long spread of medium patriarchy levels linking Catalonia and southwestern France with various culturally and geographically disparate areas of Westphalia and Tyrol, and with a long vertical axis stretching from Lithuania to Wallachia (Romania) in southeastern Europe. In contrast to the results presented in the mainstream literature, our findings indicate that areas with elevated PI values also existed in northwestern Europe, such as in the “Bible Belt” in the southwestern part of Norway, in northwestern Germany, and on the Shetland Islands.

We also observe a considerable degree of variation within countries and across the macroregions of Europe. The territories between the Baltic, the Adriatic, and the Black Seas seem to have been particularly diverse, as they appear to have encompassed areas with low levels of patriarchy (like the western and northern parts of historical Poland), as well as areas with moderate to high levels of patriarchy (like many parts of Hungary, Slovakia, and Romania). In fact, historical Poland-Lithuania (which included modern-day Poland, Lithuania, Belarus, and large parts of Ukraine) is the only historical region for which we found a combination of high-to-low patriarchy intensities, possibly indicating a transitory, intermediate pattern (Szołtysek, 2015b). Our results also show that the German territories had highly diverse PI values, ranging from very low to medium levels.

Furthermore, Figure 4 suggests that the decline in patriarchy levels may not have been continuous or linear; that is, that conservative patriarchal cultures did not necessarily evolve into “modern,” gender-egalitarian societies. For example, many regions of Sweden had higher patriarchy levels than Denmark, even though the Swedish census was taken almost a century later than the Danish census. Moreover, Iceland had much lower PI values than Norway, despite having been surveyed one hundred years earlier. Similarly, data from Germany show that the PI values improved little from the early to the late 19th century. While regions of eastern Europe are underrepresented in the data from later periods, there are no clear signs that the patriarchy levels in these regions declined drastically with the passage of time. Finally, the data for southeastern Europe indicate that Albania of the early 20th century was much more patriarchal than several populations of the Balkans in the early 19th century or earlier.

The PI and Other Family System Measures

Given the character and the geographic distribution of the PI, a considerable overlap between this measure and some common measures of historical family systems is to be expected. This assumption seems to be confirmed by our attempt to match our findings on the distribution of patriarchal features with the results of Dennison and Ogilvie (2014). These authors created a Borda ranking of European societies based on what they called “the three European Marriage Pattern criteria” (EMP): female marriage age, female celibacy, and household complexity (Dennison & Ogilvie, 2014, pp. 669-670). To compare our results with those of Dennison and Ogilvie, we started with their division of Europe into 33 societies (some of which included country subdivisions), which they obtained from 365 research studies. We then used our dataset on 266 regional populations to derive the average PI values for this division. This approach allowed us to cover 18 of the 33 societies studied by Dennison and Ogilvie. To ensure that big cities do not dominate the outcomes, we gave each region the same weight regardless of its population size in obtaining these averages. However, when interpreting the results of this comparison, it is important to note that neither the values obtained by Dennison and Ogilvie (2014) nor our values are representative in a strict statistical sense. Nevertheless, we believe that this comparison provides us with an impression of the relationship between these two measures. The scatterplot derived from this comparison is presented in Figure 5 (a regression line was added to the scatterplot for orientation). The outcomes of the analysis suggest that there is a positive relationship between the PI and the EMP rankings (Pearson’s r: .86): that is, societies with a high PI were more likely to be characterized by early female marriage age, low rates of female celibacy, and high levels of household complexity. While this finding might not be particularly surprising, it provides support for the view that the PI is a useful measure of historic cross-cultural differentials in family organizations. Such a measure is more comprehensive than the usual triad of features commonly studied in family history research: namely, age at marriage, celibacy, and household structure (Gruber & Szołtysek, 2016; Hajnal, 1982).

Figure 5.

The relationship between the PI and a Borda ranking of European societies according to the “three European Marriage Pattern criteria.”

While it is appropriate to use the PI for such purposes, it is important to note that the observed distributions of patriarchy levels may not necessarily overlap with the spatial patterning of the three main types of family systems that are commonly assigned to historic European societies: neolocal nuclear, patrilocal stem, patrilocal joint. While it has been argued that “gender bias informs [the three main types of] family systems in contingent fashion” (Skinner, 1997, p. 58), and that this bias is most pronounced in joint family systems and is least pronounced in conjugal family systems, we believe that the family system–patriarchy relationship is more complex. For example, our findings show (see Figure 3 above) that the regions in which the conjugal–neolocal family model was most prevalent (like northern France and Romania) can still be distinguished by their relative patriarchy levels, and that the European regions in which the joint family was prevalent (e.g., in Albania, Slovakia, Lithuania, central Ukraine, and Russia) had a range of patriarchal values. Although these observations are still tentative, they may invite family historians to reassess their conceptual apparatus.

The PI and Other Gender Inequality Measures

In evaluating the relevance of the PI for wider inequality studies, two potential caveats need to be addressed. First, sex-related or age-related social inequalities are frequently not limited to the realm of family, as they are in the PI, but encompass other dimensions of social life, such as politics and labor markets (see Young et al., 1994). However, as we argued above, we believe that the PI is relevant for social dimensions beyond those associated with the family, as it appears that household organization practices helped to uphold systematic forms of gender and generational biases within societies. Nonetheless, future research should compare our results with the findings of historical patriarchy studies that are based on other sources and that cover other spheres, provided such studies are conducted on a sufficiently large scale.

Second, it is important to remember that unlike most existing social science indicators of sex discrimination, the PI does not just reflect gender differences (or women’s status), but merges the gender dimension with that of seniority. While this approach provides a more comprehensive account of the multidimensionality of empowerment and agency than most other measures of patriarchy offer (see Narayan, 2006), it makes it more difficult to compare the PI with more gender-focused measures.

Given these ambiguities, it is interesting to explore the question of to what extent the variation in the combination of various historical family-related institutions and societal mechanisms that the PI captures is similar to the present-day spatial variation in macroindicators of gender inequality. We therefore decided to compare the historical variety in PI levels with today’s (2013) spatial variation based on a well-established measure from inequality research: the Gender Inequality Index (GII).⁷ While some of the GII’s components correspond roughly to certain components of the PI, they were obtained in a different manner using contemporary data. In deriving the PI values for this assessment, we followed the procedure we applied in the comparison of our data with the Dennison and Ogilvie data, except that in this case we were basing the divisions on the present-day countries for which GII data are available. As in the exercise above, we compared only those countries for which PI data are existing.

A comparison of the GII levels with our derived PI values is presented in Figure 6. Again, the comparison should be interpreted with caution, as our PI data for the present-day countries are not representative in a strict statistical sense. The graph we derived suggests that there is a rather strong positive relationship between historical patriarchy levels and the GII values (Pearson’s r: .65). Thus, it appears that areas which had relatively high patriarchy levels in the past also tend to have relatively high gender inequality levels today. Although this comparison has some limitations, we believe that it merits attention.

Figure 6.

The relationship between the PI and the Gender Inequality Index (2013).

A similar approach can be followed in comparing the PI with the HGEI (Dilli et al., 2015). The HGEI represents the most recent methodological innovation in the measurement of gender inequality worldwide (Figure 7).⁸ For our comparison, we decided to use HGEI data for the years 2000-2010, as they are the most complete. Again, we found a clear relationship between the two measures: countries with high scores on the historical patriarchy scale tend to have low levels of gender equality today (Pearson’s r: −.72).

Figure 7.

The relationship between the PI and the Historical Gender Equality Index (data from 2000).

Beyond having a purely diagnostic dimension, the finding that historical and contemporary inequality patterns are linked suggests that gender disparities persist over the long term. The mere establishment of such associations does not, of course, allow us to posit the existence of direct causal links between the past and the present. If historical patriarchy levels influenced contemporary gender inequalities, they probably did so in a path-dependent manner. But even without making strong claims about the lasting impact of historical patterns on the patterns that exist today, we believe that the observed associations are of relevance for social scientists and researchers engaged in comparing developmental levels, as these links imply that the historical context is an important consideration when analyzing the global contours of contemporary forms of social inequality.

Patriarchy in Context

Why did some historical societies have higher levels of patriarchy, while others had lower levels? Referring to medieval England, Bennett (2007) suggested that “patriarchy was an effect of many institutions,” but did not explain this observation any further (p. 78). Therborn (2004) argued that a process of “de-patriarchalization” took place, and asserted that this development was influenced by legal changes, proletarianization, and wider processes of urbanization and industrialization (pp. 17-22; similarly, Miller, 1998; Moghadam, 1992). According to Rahman and Rao (2004), the key determinants of female inequity were cultural norms (especially regarding kinship), economic conditions, and state policies and legislation. Meanwhile, Alesina hypothesized that traditional agricultural practices influenced the historical gender division of labor and the evolution of gender norms (Alesina, Giuliano, & Nunn, 2013; earlier Boserup, 1970; see also Carmichael & Rijpma, 2017).

Using our data to address the question posed above is a challenging endeavor, especially given the inherent difficulties we face in obtaining comprehensive information on potential covariates from the surviving body of historical statistics. Also, as the PI captures a multidimensional phenomenon, it is associated with a wide range of sociodemographic and cultural dimensions. Hence, any modeling attempt is likely to be confronted with multicollinearity issues. In addition, as many of the cultural and institutional traits that may influence patriarchy are likely to be influenced by patriarchy themselves, the relationship might go both ways. Thus, the investigation that follows merely represents a first attempt to explore a number of hypotheses related to conditions that are potentially relevant for understanding variation in patriarchy levels. In this analysis, we derived spatially sensitive regression estimates⁹ of the associations between patriarchy levels and a broad range of socioeconomic and institutional characteristics of the regional populations, as well as regional variation in the degree of demographic spatial centrality and environmental conditions.¹⁰

The PI serves as our dependent variable. In attempting to account for variation in living standards, we decided to include the proportion of the population who were elderly (aged 65+) and the child–woman ratio (CWR) as covariates. The latter indicator is the ratio of children under age 5 to the number of women between ages 15 and 49 (see Willigan & Lynch, 1982).¹¹ We assume that regions with a relatively high level of development also had relatively high proportions of elderly people and relatively low patriarchy levels (Rosset, 1964).¹² To interpret the CWR, we have to take endogeneity concerns into account, as the CWR might be more than just a proxy for the level of development. It is also likely that the link between patriarchy and fertility levels was positive (Dyson & Moore, 1983). Overall, however, we expect to find that in areas with relatively high levels of development, both the CWR and the patriarchy values would have been low.¹³

To explore the potentially “depatriarchalizing” role of urban (industrial) life, we included a covariate for the share of the population in each region who were living in rural areas (see Note 5). Next, to account for whether the region was more centrally or more peripherally located within Europe, we derived a “population potential” covariate (see Stewart & Warntz, 1958). This provides information of whether a specific region was situated close to important population centers of Europe, or rather in peripheral sparsely populated areas (see Online Appendix 1 for technical details). During our period of observation, the cost of transport was still an essential factor in the extent to which people had access to markets. Accordingly, the potential for economic growth was usually relatively low in peripheral areas that were situated far away from important population centers. In addition, the extent to which large numbers of people were living in close proximity was an important factor in determining the degree to which knowledge and skills were diffusing and being maintained (Goldin, 2016). Overall, we expect to find a negative association between the population potential and the PI.

A control for terrain ruggedness was included to account for variation in the opportunities for economic and social development (Jimenez-Ayora & Ulubaşoğlu, 2015; see Online Appendix 1 for technical details). Rugged topography may affect the ability of residents to engage in intensive agricultural activities, and their access to public infrastructure such as educational institutions or the transport system. This issue was of particular relevance in the period between 1850 and 1950, when having access to railways was an important determinant of the developmental prospects of a given region. Compared with roads, railways are more difficult to build in rugged terrain.

Moreover, in regions with rugged terrain cultural anomalies may persist longer as, at least in historical times, such a terrain may have constrained people to communicate with individuals in nearby locations. However, the relationship between the terrain ruggedness of a region and its development prospects is not clear-cut, as some of these areas had access to water energy and/or mineral deposits. Exploiting these resources would have provided the populations in these regions with opportunities to engage in proto-industrialization; a process that is generally associated with the depatriarchalization of family relations (Medick, 1976). Overall, however, we expect to find a positive association between terrain ruggedness and PI levels across Europe.

We also added a covariate that indicates whether the populations were subjected to serfdom. According to our considerations, there are three possible channels through which serfdom may have increased the patriarchal bias among these regions. First, the regions with serfdom tended to have more complex families that were more likely to exhibit high PI levels. Second, the Russian version of serfdom in particular provided conditions in which the authority of the household patriarch was institutionally endowed by the seigniors. Finally, because of its heavy reliance on coerced labor with draught animals (corvée), serfdom created structural conditions that devalued female labor (Alesina et al., 2013; Szołtysek, 2015b, vol. 1). Thus, we can assume that serfdom had negative effects on women’s status and agency levels.

Another important aspect that should be considered in this context is the classification according to the period in which all or most of the data for each of our 266 regional populations were collected. Based on the general consensus of the sociological and the historical literature (Therborn, 2004), we expect to find that patriarchy levels decreased over time. For our models, we considered the following categories: pre-1800, 1800-1850, and after 1850 (reference category). Finally, we included dummies for 11 regions of Europe (as depicted in Figure 1) in an attempt to account for unobserved developmental effects, such as the efficiency of the bureaucracy, the role of the labor markets, and the legal system (with Germany used as a reference category). In addition, we considered several other variables, such as the share of cropland, the rules of descent, and the dominant religion. However, for a number of reasons, we decided not to account for these variables in the analysis.¹⁴ As the included attributes are limited, the regression models should not be interpreted as an attempt to establish causality. The main purpose is to explore the associations between the PI and the available covariates in a multivariate framework.

To account for variation in the density of locations across various parts of Europe, we decided to apply weights to ensure that each of the seven large regions is given equal weight in the regressions (Balkans, East, Germany, Great Britain, Habsburg, Scandinavia, and West). As we are analyzing spatial data, it is likely that our regressions are influenced by spatial autocorrelation, which might introduce bias into both the coefficient estimates and the obtained significance levels (Bivand, Pebesma, & Gómez-Rubio, 2013). To explore the degree to which our models are affected by spatial autocorrelation, we derived the Moran’s I index of spatial autocorrelation for the model residuals. The Moran’s I is very similar to Pearson’s correlation coefficient, except that it does not control for the correlation between two variables x and y, but for the correlation between a variable y in region i, and the value of the (weighted) mean of y in neighboring regions j. In calculating the Moran’s I, we considered the five nearest neighboring regions.¹⁵ To determine whether our model estimates are potentially biased due to multicollinearity, we applied variance inflation factors (VIF).

Table 4 presents the results of our regressions. We would first like to provide an overview of the spatial clustering of the dependent variable. As we applied regional weights in the regressions, we decided not to derive the Moran’s I for the dependent variable, but instead to calculate a base model that simply takes into account the dependent variable, the intercept, and the weights. For the residuals of this model, we then derived the Moran’s I. The results of this approach show that, in line with the visual impression in Figure 3, the patriarchy levels in our dataset are characterized by extremely high levels of positive spatial autocorrelation. In Model 1, we just controlled for our demographic proxies for development levels and the time period. In this model, only the proportion of elderly people is significant, and the coefficient estimate is in the expected direction. However, the Moran’s I on the residuals indicates that there are high levels of positive spatial autocorrelation. This finding implies that the independence assumption is probably violated, which could in turn cause bias in the coefficient estimates and increase the likeliness to obtain too high significance levels.

Table 4.

Regression Results.

	Model 1					Model 2					Model 3
	β	SE	Std. β	p	VIF	β	SE	Std. β	p	VIF	β	SE	Std. β	p	VIF
(Intercept)	19.10	1.49	0.00	***	0.00	12.20	1.25	0.00	***	0.00	8.46	0.81	0.00	***	0.00
Proportion of elderly (65+ years)	−1.09	0.13	−0.38	***	1.21	−0.82	0.12	−0.28	***	1.68	0.07	0.06	0.02		2.45
Child–woman ratio (see also note 11)	2.97	2.12	0.06		1.18	6.18	1.81	0.12	***	1.30	5.00	1.09	0.10	***	2.41
Share rural						1.07	0.59	0.07	^†	1.24	2.21	0.29	0.14	***	1.74
Population potential (ref.: High)
Low						1.23	0.45	0.12	**	1.42	1.94	0.45	0.18	***	6.91
Medium						−0.53	0.41	−0.05		1.33	−0.19	0.28	−0.02		2.83
Terrain ruggedness (log-transformed)						0.43	0.14	0.10	**	1.27	−0.05	0.09	−0.01		2.81
Serfdom (yes)						3.33	0.65	0.17	***	1.86	0.30	0.55	0.02		5.90
Period (ref.: after 1850)
Before 1800	0.83	0.62	0.06		1.17	1.54	0.53	0.11	**	1.61	4.88	0.73	0.36	***	13.88
1800-1850	0.77	0.51	0.07		1.14	2.28	0.40	0.20	***	1.35	0.87	0.55	0.08		10.38
Region (ref.: Germany)
Albania (Balkans)											13.90	0.74	0.63	***	5.61
Southeast (Balkans)											10.85	0.66	0.47	***	4.89
Central-East (East)											−1.11	0.74	−0.03		4.01
East (East)											9.55	0.92	0.37	***	12.07
England (Great Britain)											−1.52	0.60	−0.11	*	4.21
Scotland (Great Britain)											−2.64	0.61	−0.17	***	3.59
Wales (Great Britain)											−2.79	0.73	−0.12	***	1.97
Habsburg											5.24	0.58	0.24	***	5.60
Scandinavia											−2.08	0.68	−0.18	**	10.47
West											7.86	0.65	0.36	***	7.88
Interaction effects: See below
N	264					264					264
Adjusted r²	.10					.42					.92
Moran’s I (with p value)
Residuals of base model	0.89***					0.89***					0.89***
Residuals	0.81***					0.81***					0.03**
Interaction effects
Southeast (Balkans)before 1800											−7.77	1.35	−0.10	***	1.82
Southeast (Balkans)1800-1850											−6.30	0.92	−0.16	***	3.33
East (East)before 1800											−5.51	1.24	−0.15	***	12.22
East (East)1800-1850											−2.73	1.20	−0.07	*	8.16
Habsburgbefore 1800											−3.65	1.21	−0.05	**	2.79
Scandinaviabefore 1800											−4.59	0.87	−0.26	***	6.51
Scandinavia1800-1850											2.34	0.75	0.12	**	4.13
West1800-1850											−7.57	0.80	−0.29	***	8.53

Note. β = coefficient estimate; SE = standard error; Std. β = standardized coefficient estimate; VIF = variance inflation factor.

†

p < .1. *p < .05. **p < .01. ***p < .001.

In Model 2, we added all the other socioeconomic and environmental covariates. This substantially increased the r², but it did not reduce the high degree of spatial autocorrelation among the residuals. Thus, the estimates of Model 2 also have to be interpreted with caution. Compared with the outcomes of Model 1, the estimated coefficient for the elderly is attenuated, but remains highly significant. The CWR and the time dummies are significant in the expected direction. However, the time dummies do not display the expected negative gradient. The outcomes for all the other significant variables are in the anticipated direction. Relative to other areas, the regions with serfdom had significantly higher patriarchy levels, and those with a high degree of terrain ruggedness had higher PIs. In addition, peripheral areas with low population potential were significantly more likely than the reference group (regions with high population potential) to have had high PI levels. The outcome for the variable capturing the share of the population living in rural areas is significant at the .1 level, and indicates that rural areas had higher PI levels.

To further reduce the spatial autocorrelation in our models, we included our regional controls in Model 3. Furthermore, because observations for only one or two periods were available for a number of our 11 macroregions (which made tracking changes over time within these regions very difficult),¹⁶ we also added interaction effects between the regions and the time periods. The introduction of these regional dummies and the interaction effects allowed us to reduce the Moran’s I on the residuals substantially, to 0.03. Although this value is still significant at the .01 level, it provides us with some confidence that Model 3 is much less biased due to spatial autocorrelation. However, introducing the regional dummies came at a price: namely, that the dummies might act as a proxy for some of the other covariates. Our outcomes for serfdom are likely affected by this problem, as the areas with serfdom were primarily concentrated in the region East (East). Thus, it is not surprising that the coefficient for serfdom is not significant in the full model, while we obtained highly significant positive estimates for our dummy East (East).This finding is further corroborated by the VIF values, which are high for these two variables in Model 3.

We refrain from discussing the interaction effects, as they are in part difficult to interpret. Of the other covariates, all the significant variables are in the expected direction. In Model 3, we also obtained the expected negative gradient for the time dummies. However, the VIFs for these two dummies are very high. Terrain ruggedness is no longer significant, possibly due to the fact that a large share of the populations who lived in rugged terrain is clustered in certain regions (e.g., in Albania). However, the VIF for terrain ruggedness is not high. Regional dummies seem to account for a large portion of the variation. It can also be noted that the share of the population living in rural areas and the population potential variable appear to be more relevant for understanding variation in patriarchy levels in Model 3 than in Model 2. The share of the elderly in the population, on the other hand, is not significant in Model 3.

When we excluded the CWR from Model 3 based on endogeneity concerns, the outcomes for the other variables did not change substantially. We also ran separate models on the Mosaic and the NAPP regions. The outcomes of these models differ especially with regard to the population potential variable. There is a much more pronounced association between high PI values and low population potential in the Mosaic than in the NAPP dataset. A possible explanation for this result is that remoteness might have played a much bigger role in areas located in the central part of the continent compared with Great Britain and Scandinavia, as this latter group of regions likely had easier access to the sea/international trade.

While our modeling attempts were subject to a number of limitations, the outcomes provide us with some confidence that there was a negative association between development levels and PI levels in historical Europe. In particular, our findings suggest that patriarchy levels in the past were especially high in rural and peripherally located areas. The role of geocovariates for understanding variation seems to be particularly pronounced for continental Europe other than Scandinavia.

Conclusion

In this study, we sought to move the analysis of historical trends in gender inequality beyond the usual confines of a one-dimensional focus on sex-stratification/discrimination. With the PI, we proposed a historical inequality measure that combines the power of the father and the power of the husband with other dimensions in a composite approach. By doing this, we demonstrated that limited but widely available historical data can be used to construct variables that allow to measure historic trends in gender and generational relations across Europe. Moreover, we presented the argument that by comprising localized indicators that combine both gender and age relations, the index allows to better account for the historical cross-cutting of gender bias with other forms of discrimination. We believe that these contributions will facilitate the historical reconstruction of the dynamics of power in preindustrial Europe, and enhance the current body of historical statistics on cross-societal inequalities.

We applied the PI to census microdata to provide an account of the regional prevalence of gender- and age-based authority patterns across Europe with a focus on the 18th and 19th centuries. This analysis showed that the complex societies of (western) Eurasia (Goody, 1976) differed significantly in their patriarchy levels as conceptualized in the PI. As the spatial contours of this variation do not necessarily align with the corresponding spatial patterning of the main types of historical family systems, family historians may wish to further explore this line of research.

Our finding that the historical PI values are associated with values obtained by contemporary measures of gender inequality provides support for the argument that variation in historical conditions, structures, and institutions can be relevant for understanding contemporary spatial disparities in development, well-being, and wealth (e.g., Nunn, 2009). It also reiterates the importance of the family and the household as historically crucial sites for generating societal inequalities (Alesina et al., 2013).

Our regression results suggest that PI values tended to be higher in areas that were more remote and less well integrated. Can we assume that the more diversified social structures of cities and densely populated regions mitigated against higher PI levels? Or should we interpret elevated PI values as adaptation mechanisms that were triggered in response to challenges and constraints created by local geographic externalities? In other words, can these higher PI values be seen as having arisen in response to low levels of state penetration, weak institutions, and poor access to public services and infrastructure; and hence in reaction to broader “spatial poverty traps” (Bird, Higgins, & Harris, 2010)? Future research should attempt to explore these issues further. It is equally important that scholars continue seeking to better understand the relationship between patriarchal structures and prospects for human development. The Mosaic and the NAPP data could be very well suited to that line of research.

Footnotes

Acknowledgements

The authors acknowledge comments from Patrick Heady, Auke Rijpma, and Jan Kok. They thank Julia Szołtysek and Miriam Hils for language editing.

Authors’ Note

The maps used in this publication are partly based on the following source: ©EuroGeographics for the administrative boundaries.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Mikołaj Szołtysek is a senior research fellow at the Max Planck Institute for Social Anthropology in Halle/Saale. He was one of the founders of Mosaic—one of the largest data infrastructure projects in historical demography for research on family patterns in historic Europe.

Sebastian Klüsener is deputy head of the Laboratory of Fertility and Well-Being at the Max Planck Institute for Demographic Research in Rostock, Germany. He holds a PhD in geography, and one of his main research interests is to study spatial aspects of long-term social change processes.

Radosław Poniat works at the department of History and Political Science at the University of Białystok. His research interest focuses on the social history and the sociology of the social structures in historical eastern Europe.

Siegfried Gruber has earned his doctoral degree at the University of Graz, where he is affiliated to the Department of History. His main research interests focus on the spatial variation of European household patterns and patriarchal structures (especially in southeastern Europe).

References

Alesina

Giuliano

(2014). Family ties. In Aghion

Durlauf

S. N.

(Eds.), Handbook of economic growth (pp. 177-215). Amsterdam, The Netherlands: Elsevier North Holland.

Alesina

Giuliano

Nunn

(2013). On the origins of gender roles: Women and the plough. The Quarterly Journal of Economics, 128, 469-530.

Bennett

J. M.

(2007). History matters: Patriarchy and the challenge of feminism. Philadelphia, PA: University of Pennsylvania Press.

Bird

Higgins

Harris

(2010). Spatial poverty traps: An overview (CPRC Working Paper 161). London, England: Chronic Poverty Research Centre.

Bivand

R. S.

Pebesma

Gómez-Rubio

(2013). Applied spatial data analysis with R (2nd ed.). New York, NY: Springer.

Boserup

(1970). Woman’s role in economic development. London, England: George Allen & Unwin.

Carmichael

Dilli

Rijpma

(2014). Gender inequality in a long term perspective. In van Zanden

J. L.

Mira d’Ercole

Rijpma

Smith

Timmer

(Eds.), Global well-being and development: A long-term perspective since 1820 (pp. 217-248). Paris, France: Organisation for Economic Co-operation and Development.

Carmichael

Dilli

van Zanden

J. L.

(2016). Introduction: Family systems and economic development. Economic History of Developing Regions, 31, 1-9.

Carmichael

Rijpma

. (2017). Blood is thicker than water: Geography and the dispersal of family characteristics across the globe. Cross-Cultural Research, 51, 142–171.doi: 10.1177/1069397117691025

10.

Coltrane

Adams

(2000). Gender and families. Lanham, MD: AltaMira Press.

11.

Dennison

Ogilvie

(2014). Does the European Marriage Pattern explain economic growth? The Journal of Economic History, 74, 651-693.

12.

Dilli

Rijpma

Carmichael

(2015). Achieving gender equality: Development versus historical legacies. CESifo Economic Studies, 61, 301-334.

13.

Dorius

S. F.

Firebaugh

(2010). Trends in global gender inequality. Social Forces, 88, 1941-1968.

14.

Drwenski

(2015). Data on inequality and the inequality of data: The last two centuries. Journal of World-Historical Information, 2-2(1). doi:10.5195/jwhi.2015.15

15.

Dyson

Moore

(1983). On kinship structure, female autonomy, and demographic behavior in India. Population and Development Review, 9, 35-60.

16.

Folbre

(1986). Cleaning house: New perspectives on households and economic development. Journal of Development Economics, 22, 5-40.

17.

Goldin

(2016). Human capital. In Diebolt

Haupert

(Eds.), Handbook of cliometrics (pp. 55-86). Berlin and Heidelberg, Germany: Springer.

18.

Goody

(1976). Production and reproduction: A comparative study of the domestic domain. Cambridge, England: Cambridge University Press.

19.

Gruber

Szołtysek

(2016). The patriarchy index: A comparative study of power relations across historical Europe. The History of the Family, 21, 133-174.

20.

Hajnal

(1982). Two kinds of preindustrial household formation system. Population and Development Review, 8, 449-494.

21.

Halpern

J. M.

Kaser

Wagner

R. A.

(1996). Patriarchy in the Balkans: Temporal and cross-cultural approaches. The History of the Family, 1, 425-442.

22.

Jimenez-Ayora

Ulubaşoğlu

M. A.

(2015). What underlies weak states? The role of terrain ruggedness. European Journal of Political Economy, 39, 167-183.

23.

Johnston

D. F.

(1985). The development of social statistics and indicators on the status of women. Social Indicators Research, 16, 233-261.

24.

Joseph

(1996). Patriarchy and development in the Arab world. Gender & Development, 4, 14-19.

25.

Kandiyoti

(1988). Bargaining with patriarchy. Gender & Society, 2, 274-290.

26.

Klasen

(2006). UNDP’s gender-related measures: Some conceptual problems and possible solutions. Journal of Human Development, 7, 243-274.

27.

Klüsener

Devos

Ekamper

Gregory

I. N.

Gruber

Martí-Henneberg

. . . Solli

(2014). Spatial inequalities in infant survival at an early stage of the longevity revolution: A pan-European view across 5000+ regions and localities in 1910. Demographic Research, 30, 1849-1864.

28.

Kok

(2014, December 18-20). Women’s agency in historical family systems. Paper presented at the conference “The Deep Causes of Economic Development,” Utrecht, The Netherlands. Retrieved from http://www.cgeh.nl/deep-causes-economic-development

29.

Koller

Stahel

W. A.

(2011). Sharpening Wald-type inference in robust regression for small samples. Computational Statistics & Data Analysis, 55, 2504-2515.

30.

Malhotra

Sidney

R. S.

Boender

(2002). Measuring women’s empowerment as a variable in international development. Washington, DC: International Center for Research on Women and the Gender and Development Group of the World Bank, The World Bank.

31.

Malhotra

Vanneman

Kishor

(1995). Fertility, dimensions of patriarchy, and development in India. Population and Development Review, 21, 281-305.

32.

Mason

K. O.

(1986). The status of women: Conceptual and methodological issues in demographic studies. Sociological Forum, 1, 284-300.

33.

Medick

(1976). The proto-industrial family economy: The structural function of household and family during the transition from peasant society to industrial capitalism. Social History, 1, 291-315.

34.

Milanovic

(2005). Worlds apart: Measuring global and international inequality. Princeton, NJ: Princeton University Press.

35.

Milanovic

Lindert

P. H.

Williamson

J. G.

(2011). Pre-industrial inequality. The Economic Journal, 121, 255-272.

36.

Miller

(1998). Transformations of patriarchy in the west, 1500-1900. Bloomington, IN: Indiana University Press.

37.

Moghadam

V. M.

(1992). Patriarchy and the politics of gender in modernizing societies: Iran, Pakistan and Afghanistan. International Sociology, 7, 35-53.

38.

Murdock

G. P.

(1949). Social structure. New York, NY: Macmillan.

39.

Narayan

. (Ed.). (2006). Measuring empowerment: Cross-disciplinary perspectives. Washington, DC: The World Bank.

40.

Nunn

(2009). The importance of history for economic development. Annual Review of Economics, 1, 65-92.

41.

Rahman

Rao

(2004). The determinants of gender equity in India: Examining Dyson and Moore’s thesis with new data. Population and Development Review, 30, 239-268.

42.

Reher

D. S.

(1998). Family ties in Western Europe: Persistent contrasts. Population and Development Review, 4, 203-234.

43.

Rosset

(1964). Aging process of population. New York, NY: Macmillan.

44.

Ruggles

Roberts

Sarkar

Sobek

(2011). The North Atlantic Population Project: Progress and prospects. Historical Methods, 44, 1-6.

45.

Schlegel

(1972). Male dominance and female autonomy. New Haven, CT: HRAF Press.

46.

Skinner

G. W.

(1997). Family systems and demographic processes. In Kertzer

D. I.

Fricke

(Eds.), Anthropological demography: Toward a new synthesis (pp. 53-95). Chicago, IL: University of Chicago Press.

47.

Stewart

J. Q.

Warntz

(1958). Macrogeography and social science. Geographical Review, 48, 167-184.

48.

Szołtysek

(2015a). Households and family systems in early modern Europe. In Scott

(Ed.), The Oxford Handbook of Early Modern European History, 1350-1750: Vol. I. Peoples and place (pp. 313-341). Oxford, England: Oxford University Press.

49.

Szołtysek

(2015b). Rethinking East-central Europe: Family systems and co-residence in the Polish-Lithuanian Commonwealth. Bern, Switzerland: Peter Lang.

50.

Szołtysek

Gruber

(2016). Mosaic: recovering surviving census records and reconstructing the familial history of Europe. The History of the Family, 21(1), 38-60.

51.

Therborn

(2004). Between sex and power: Family in the world, 1900-2000. London, England: Routledge.

52.

Therborn

(Ed.). (2006). Inequalities of the world: New theoretical frameworks, multiple empirical approaches. London, England: Verso.

53.

Todd

(1987). The causes of progress: Culture, authority and change. New York, NY: Basil.

54.

van Zanden

J. L.

Baten

Foldvari

van Leeuwen

. (2014). The changing shape of global inequality 1820-2000; exploring a new dataset. Review of Income and Wealth, 60, 279-297.

55.

van Zanden

J. L.

Baten

Mira d’Ercole

Rijpma

Smith

Timmer

. (Eds.). (2014). How was life? Global well-being since 1820. Paris, France: Organisation for Economic Co-operation and Development.

56.

Whyte

M. K.

(1978). The status of women in pre-industrial societies. Princeton, NJ: Princeton University Press.

57.

Willigan

J. D.

Lynch

K. A.

(1982). Sources and methods of historical demography. New York, NY: Academic Press.

58.

World Bank. (2011). World Development Report 2012: Gender equality and development. Washington, DC: Author.

59.

Yohai

V. J.

(1987). High breakdown-point and high efficiency robust estimates for regression. The Annals of Statistics, 15, 642-656.

60.

Young

Fort

Danner

(1994). Moving from the “status of women” to “gender inequality”: Conceptualisation, social indicators and an empirical application. International Sociology, 9, 55-85.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.19 MB