Table 1.
Characteristics of the samples.
Fig 1.
Subnational data distribution by macro-regions.
Notes: Data source and regional divisions as presented in Table 1. Country and subnational boundaries were derived from public domain data provided by Natural Earth (https://www.naturalearthdata.com) via the rnaturalearth R package [59]. Data are in the public domain and therefore compatible with the CC BY 4.0 license.
Fig 2.
The four major dimensions of patriarchy and their input variables.
Notes: Indicators (from top to bottom): (1) female household heads aged 20+ among all adult heads, (2) ever-married women aged 15–19, (3) couples in which the wife is older than the husband, (4) women aged 20–24 living only with non-kin, (5) men aged 65+ in households headed by a younger male relative, (6) male heads aged 20–29 living only with their immediate family, (7) persons aged 65 + living with at least one lateral kin, (8) persons aged 65 + living without relatives or a spouse, (9) ever-married women aged 15–30 living with an adult male kin of the husband or his mother, (10) boys among children aged 10–14, and (11) child sex ratio (boys to girls) aged 0–4. For details on the measurement of these components and IPUMS-based adjustments, see Table A in S1 Text. (+/−) indicates the expected direction of each component’s relationship with societal patriarchy levels. Arrows denote the suggested relationships between constructs in the model: reflective lower-order constructs and the formative higher-order construct.
Fig 3.
Correlation matrix of the four sub-indices of patriarchy (Kendall’s Tau).
Notes: Data source as in Table 1. All colour quadrants indicate significant correlation (p-value<0.05).
Fig 4.
Country-level correlations of PI and its components with main gender (in)equality indices.
Notes: Data source for PI and its four domains, as in Table 1. For other indices (in order of appearance): (GEM) Gender Empowerment Measure – values are from the closest available date to the census year, i.e., 1995, 2006 (India), or 2009, 2011 (Cambodia; higher value (closer to 1) = greater empowerment of women); see [146–148]. (GGGI) Global Gender Gap Index – values are from the closest available date to the census year – i.e., 2006, 2011, 2019, 2020, 2022; higher score (closer to 1.00) = more gender parity) [149]. (HGEI) Historical Gender Equality Index – the closest available date (1997–2003) to the census year (higher score = more equality) [150]. (GDI) Gender Development Index – exact census years (values below 1 = higher human development for men than women; values above 1 = the opposite) [151]. The gap between HDI and GDI (expressed as a difference) is used following [152], where it is meant to serve as a useful proxy for the developmental costs associated with gender inequality. (GII) Gender Inequality Index – exact census years (higher score = lower inequality between women and men) [153]. (SIGI) Social Institutions and Gender Index – values are from the closest available date to the census year (i.e., 2009, 2012, 2014; higher score = more gender-based discrimination) [154]. (SIGI_FamCode) Discriminatory Family Code component of the SIGI – values are from the closest available date to the census year (i.e., 2009, 2012, 2014; higher score = more discrimination) [154]. (SIGI_PhysIntegr) Restricted Physical Integrity component of the SIGI – all values from 2019 (higher score = more restriction) [154].
Table 2.
Correlations of PI with available subnational measures of gender (in)equality.
Fig 5.
Country-level distribution of PI.
Notes: Data source as in Table 1.
Fig 6.
Notes: Data source and regional divisions as presented in Table 1. Boundaries: Natural Earth, public domain [59].
Fig 7.
Within-country variation in PI.
Notes: Data source and regional divisions as presented in Table 1.
Fig 8.
LISA significant (p < 0.001) clusters for the distribution of the Asian PI.
Notes: Data source and regional divisions as presented in Table 1. Boundaries: Natural Earth, public domain [59].
Table 3.
Theil’s inequality measures in PI distribution across Asia.
Fig 9.
Estimated effects of PI and its subcomponents on selected gender (in)equality outcomes (OLS).
Notes: The variation in the number of observations across models stems from differences in data availability. In some cases, certain regions or entire countries were excluded due to missing or non-harmonized data, or because available data referred to time periods too distant from the reference year. Additionally, some variables were only reported at broader territorial levels than others, limiting consistent coverage across all units. For full model results, see S1 Tables. Response variables: FLFP – relative female labor force participation (absolute_female/ absolute_men); exact census years as PI. Expected Schooling Ratio – Female/Male ratio in expected years of schooling (EYS), where EYS is defined as the number of years of schooling a child of school entrance age can expect to receive, if prevailing patterns of age-specific enrolment rates persist throughout the child’s schooling life. Values are from the closest available date to the census. Wtot ratio – the Female/Male ratio in age-heaping as measured by the Total Modified Whipple’s Index (Wtot) [178]. CGFR – cross gender friendship ratio; the share of female friends among male users on Facebook divided by the share of female friends among female users in a given location. The closer the value of the ratio to 1 the more men and women form equal shares of their FB ties with women. The data used here reflects friendships as of January 30, 2025. The available data cover 348 subnational regions from our collection, with approximately 75% drawn from Philippines, Turkey, Vietnam, India, and Indonesia. Countries where Facebook is inaccessible (e.g., China, Iran) are not included [177]. Life expectancy at birth(e0) gap – the relative gender gap in e0 with respect to men’s e0 (Women’s e0– Men’s e0)/Men’s e0) [175]. Values are from the closest available date to the census. Control variables: % urban – proportion of the households in a region considered as located in a place labelled as urban or rural. GNI – Log Gross National Income per capita in thousands of US Dollars (2011 PPP). Sources: for FLFP, Wtot and % urban – computed, respectively, from IPUMS-I variables LABFORCE, URBAN, and AGE for the same censuses as the PI; for EYS ratio, e0 gap, see [179]. % urban for Mongolia 2000, see [180].