Multiple scripts, multiple institutions: Introducing complexity into the understanding of women’s empowerment

Abstract

Partial enactment of women’s rights is at the crux of this analysis, which identifies factors associated with the adoption of some global women’s rights scripts but not others. Women who partially enact global principles are an important group, and focusing on them provides clues into when, where, and how institutionalized scripts are in competition. To explore this issue, Demographic and Health Survey data from 25 low- and middle-income countries across two time periods are used, with a focus on two dimensions of women’s empowerment: a woman’s household decision-making power and her attitudes toward intimate partner violence. Multinomial regressions reveal that exposure to global culture is associated with dual enactment of the two dimensions. Among partial adopters, enactment privileging physical integrity is mediated through local community institutions, including religions, whereas partial-enactment privileging decision making is associated with women’s household bargaining power.

Introduction

In this article, we address complexities within global scripts by focusing on multiple dimensions of women’s rights simultaneously. As Charles (2020) notes, “comparative scholars often treat gender ideology as a unidimensional entity, whose diverse indicators rise and fall together in response to increasing societal egalitarianism or traditionalism” (p. 87). A woman’s enactment of a woman’s rights script along one dimension does not necessarily correspond with her enactment along all dimensions. We take partial enactment of women’s rights seriously and are interested in identifying the factors explaining the embrace of some global women’s rights scripts but not others. Women who partially enact global scripts are an important group, but have not previously been the focus of research on rights. These women provide clues into when, where, and how institutionalized scripts are in competition, and the implications for the current level of adoption of women’s rights in the global system.

The content of global scripts is neither static nor homogenic. Variation in acceptance, often expressed through the idea of decoupling, can be the result of local culture and politics (Pope and Meyer, 2016), or contextualized as glocalization, the merging of the local and the global (Drori et al., 2014). Overlapping frames can lead to disagreement (Boyle et al., 2015), knowledge and institutions may change over time (Barrett and Frank, 1999; Rotem, 2022; Ulybina, 2023), and competing institutions promote different ends (Bromley et al., 2020; Hadler and Symons, 2018; Lerch et al., 2022). Embracing the complexity is especially important in the burgeoning stream of research focused on the adoption of global scripts by individuals worldwide (e.g. Charles, 2020; Hadler, 2017). For example, the issue of “script resonance” (Ozgen and Koenig, 2022), that is, the fitting of global knowledge with local societies, is complicated by the plurality of institutions at play. Such messiness in global knowledge is often ignored by research, but addressing it provides a more accurate representation of the social world.

With this analysis, we highlight that global scripts are not made of one piece and should be treated as multi-dimensional phenomena. Consequently, we examine the adoption of different elements of a global script by individuals. We use Demographic and Health Survey (DHS) data from 25 countries across two time periods (Boyle et al., 2022) to explore partial enactment of women’s rights scripts. We focus on two dimensions of women’s empowerment: a woman’s household decision-making power and her attitudes toward intimate partner violence (IPV). Combining the two dimensions leads to the following four categories: (1) dual enactment, (2) partial-enactment privileging household decision making, (3) partial-enactment privileging opposition to IPV, and (4) no enactment. Descriptive statistics show the persistence of each category over time, and multinomial logistic analyses point to the factors that are associated with each form of enactment.

To preview our findings, women more embedded in global culture—these who consume media, with more education, and urban dwellers—are more likely to be dual enactors. In terms of the partial-enactment categories, local institutions, measured through individual and communal religious affiliations, provide a strong indicator. Muslim women are more likely to reject IPV while holding minimal household decision-making power, while Catholic women and women with more household bargaining power are more likely to view IPV favorably while wielding more decision-making power. Institutions related to global scripts, to local culture and norms, and resource distribution are all at work influencing the way women embrace different possibilities of gender equity.

Theoretical background

The goal of this research is to better understand women who operate in a manner that is partially but not wholly consistent with the global women’s rights discourses promoted by international organizations. World society theory addresses the diffusion of global principles and therefore provides a useful starting point for this analysis. The theory is a constructivist approach to globalization, highlighting institutionalized norms and discourses that are shared across nation-states, international governmental and nongovernmental organizations (IGOs and INGOs), professionals, and other global actors (Boli and Thomas, 1999; Meyer et al., 1997). Principles presumed to be universal, for example, human rights, emerge from this system and become institutionalized in international and national laws, policies, and bureaucracies (Hafner-Burton and Tsutsui, 2005). The global diffusion of shared values, such as women’s rights, across an incredibly diverse range of states, is evidence of the power of global norms (Ramirez et al., 1997). Our research adds to a growing segment of world society research that moves beyond nation-state actions to consider the relationship between global norms and individuals (Boyle, 2002; Charles, 2020; Hadler, 2017; Pandian, 2019; Pierotti, 2013).

From a world society perspective, women’s rights have come to constitute a “social ether” (Drori and Krücken, 2009: 19), their appropriateness taken for granted within the bureaucracies of international organizations. For example, in 2015, all countries in the United Nations agreed on 17 Sustainable Development Goals (SDGs)—serving as moral and practical milestones for humanity—to be attained by 2030. The fifth SDG speaks directly to women’s rights, with the goal to “achieve gender equality and empower all women and girls” and is the current iteration of a process of institutionalization for women’s rights that began decades ago (Berkovitch, 1999). Thus, women’s empowerment is a core component of the global scripts surrounding women’s rights (see Desai et al., 2022). A widely endorsed definition of empowerment is the process by which those who were once denied the ability to make choices gain such ability (Kabeer, 1999). Within this broad conceptualization, women’s empowerment can have many different meanings (Eyben and Napier-Moore, 2009). For example, empowerment can be about increased opportunities, participation in decision making in public or household domains, or freedom from violence (Brysk, 2018; Moghadam and Senftova, 2005). SDG 5 attempts to capture these nuances by identifying five components of empowerment (see Figure 1).

Figure 1. The global women’s rights script: Components of Sustainable Development Goal 5.

Source: United Nations, Department of Economic and Social Affairs (see https://sdgs.un.org/goals).

To date, world society research on women’s rights has tended to focus on rights generally, for example, by using national-level quantitative human rights scores as the outcome variable (e.g. Cole, 2013; Paxton et al., 2006), or on a single aspect of women’s rights, such as representation of women in national parliaments (Swiss, 2009), abortion liberalization (Boyle et al., 2015), or the percentage of women using contraception within countries (Cole and Geist, 2021). The small but growing body of work that examines attitudinal diffusion also focuses on single outcomes, such as intimate partner violence (Pierotti, 2013) or gender equality (Charles, 2020; Kim, 2020). While illuminating, this type of research does not identify or explain the relative levels of support for different dimensions of women’s rights or the partial enactment of the array of rights by many women. Our research fills this gap by studying two very different dimensions of women’s rights simultaneously—attitudes toward IPV and a woman’s contributions to major household decisions.

Institutional contradictions

World society institutions are influential, but their enactment everywhere is not a given. They constitute only one piece of the cultural material that individuals draw on to create meaning (Friedland and Alford, 1991; Jepperson and Meyer, 2021). Every individual is embedded in a range of social contexts, such as religious, social and political groups; behavioral expectations within the areas where they reside; and the culture of the organizations where they work or attend school. Each of these overlapping normative communities has a unique set of institutionalized beliefs and practices, which can reinforce, complement, or contradict one another (Boyle, 2002; Friedland and Alford, 1991; Halliday and Carruthers, 2009). These institutions with their epistemes constrain individuals, but their gaps and inconsistencies provide opportunities for individual choice and strategic action (Alasuutari and Qadir, 2019). In most communities, some components of global discourse will coincide with the history and perspectives of the community while others will not (Ozgen and Koenig, 2022), leading to variation in acceptance, resistance, or adaptation to world society norms (Boyle, 2002; Drori et al., 2014; Pope and Meyer, 2016; Rotem, 2022). To capture this institutional complexity, we consider multiple components of global norms simultaneously.

Gender relations are the most relevant institutional frameworks for this analysis. The international system’s conceptualization of gender is linked to human rights and individualism, emphasizing women apart from the roles they play in corporate bodies (e.g. as mothers within families) and viewing them as existentially equal to each other and to men across collective boundaries (Berkovitch, 1999; Boyle et al., 2015; Frank et al., 2010). This orientation to gender relations, considered “gender equality,” appears in international law and is promoted by many educational institutions.

An alternative conceptualization is “gender essentialism,” which is less individualistic, emphasizing men’s and women’s unique responsibilities within the collective (Chatillon et al., 2018). For women, this typically means bearing and raising children. When gender essentialism is institutionalized, individuals will tend to support the idea that a healthy community emerges when men and women carry out their unique gender roles. Gender relations of this type are often associated with certain religious traditions, such as Catholicism (Boyle et al., 2017) and Islam (Kim, 2020; Mahmood, 2005), but the association with particular religious traditions varies spatially (see Cole, 2020). When gender essentialism is institutionalized within a community, men tend to be the heads of households and household decision-makers. Communities that emphasize gender essentialism may be, but are not necessarily, patriarchal (e.g. Fuseini and Kalule-Sabiti, 2015). Gender equality and gender essentialism are just two examples of the many ways gender relations can be institutionalized within communities (e.g. Knight and Brinton, 2017). Aspects of women’s empowerment promoted by international organizations will vary in their consistency with local institutions, including institutionalized gender norms; some dimensions will encounter more resistance than others.

Gender relations are not the only relevant institution for this analysis. Variation in the social construction of the individual is also important. For example, Cole (2016) in a study of 148 countries from 1980 to 2010 found more consistency in government respect for physical integrity rights compared to respect for civil liberties. Cole theorizes that norms protecting humans as biological beings (i.e. protection from physical harm) are more widely shared, and spread around the world more readily, than norms protecting humans as individual, autonomous actors, such as norms related to civil rights. Cole (2016: 721) explains, “Individuals’ capacity for pain and suffering is thought to be universal, but conceptions of the bounded and autonomous actor are culturally constructed and hence variable across cultures.” Cole’s work at the national level is consistent with other studies focused on social movements (Keck and Sikkink, 1998) and individuals (Boyle and Carbone-Lopez, 2006). Together, studies such as these suggest that opposition to IPV, a key dimension of women’s rights promoted by international organizations, should resonate widely across communities because of its close connection to protecting humans from physical harm. The implications of this work for enabling women as household decision-makers have yet to be explored.

Partial enactment of global women’s empowerment scripts

Because some components of women’s rights will likely coincide with local history and institutions while others will not (Lopes et al., 2021), it is important to consider different dimensions of women’s rights simultaneously. This allows for a better understanding of the meaning of women’s rights in local contexts. To this end, we focus on two aspects of women’s empowerment concurrently. We choose these two aspects because each corresponds to a unique dimension of SDG 5 and has been measured widely across many countries. The first is women’s participation in household decisions, which reflects the SDG 5 objective to “Recognize and value . . . the promotion of shared responsibility within the household and the family.” In our conceptualization of this type of empowerment, we answer calls to capture household dynamics (Cueva Beteta, 2006; Moghadam and Senftova, 2005) without presuming contentious relations between marital partners (Cornwall and Rivas, 2015). In this study, women’s decision making constitutes empowerment if it is either autonomous or joint with their partners. The second dimension of empowerment in this analysis is women’s attitudes toward IPV. This reflects another aspect of SDG 5, namely, the call to eliminate all forms of violence against women and girls in public and private spheres. From a rights perspective, IPV is a sign of a man’s authority over a woman, and therefore runs counter to the egalitarianism inherent in a women’s rights approach. These two dimensions address the right to autonomy over one’s body and the ability to exercise meaningful control over important decisions affecting one’s life.

Table 1 reflects the possible relationships between the two elements of empowerment for individual women. Women who are decision-makers in their households and reject IPV are represented by Cell 1; they are “dual enactors” who conform to global principles on both dimensions. The dual enactment category represents the aspirations of the global women’s rights regime. Cell 4 (“no enactment”) is the opposite of Cell 1 and represents women who have little power in their households and support IPV.

Table 1.

Women’s orientation on two dimensions of women’s empowerment.

		Women’s physical integrity rights
		Woman supports physical Integrity	Woman accepts IPV in some Cases
Women’s household decision-making power	Woman participates in household decisions	1 Decision maker opposed to IPV (dual enactors)	2 Household decision-maker who accepts IPV (partial-enactment privileging decision making)
	Woman minimally or not involved in household decisions	3 Opposes IPV with little household authority (partial-enactment privileging opposition to IPV)	4 Woman with little household authority who accepts IPV (no enactment)

Cells 2 and 3 represent partial adoption of women’s rights. In both cases, individuals reflect one principle promoted by the global system but not the other. Women who make household decisions while accepting IPV are represented by Cell 2, which we refer to as “privileging decision-making.” Women who reject IPV while minimally involved in household decision making are represented by Cell 3, which we call “privileging opposition to IPV.” Women with these combinations of characteristics are typically not the focus of studies, perhaps because there is a presumption that such combinations are only transitory stops on the way to becoming dual enactors. As we will show, this presumption is problematic, and it is therefore important to understand the women represented by these cells in their own right—who they are and the factors influencing them. Accordingly, in our analysis, we are particularly interested in the women represented by these cells.

Turning to the factors that may be associated with how women are distributed across the cells in Table 1, exposure or amenability to the principles of the global system is an important consideration. At the individual level, there are many avenues through which women may learn about and endorse globalized principles of women’s rights; access to these varies spatially (Pope and Meyer, 2016). We anticipate that these factors will be associated with women’s enactment of both dimensions of women’s rights studied here. In other words, women with these characteristics will be more likely to be dual enactors. Previous world society research on individual attitudes toward IPV (Behrman and Frye, 2021; Pierotti, 2013) and gender attitudes more generally (Charles, 2020; Kim, 2020; Pandian, 2019) has consistently found that education systems are carriers of global women’s rights messages (see also Nakagawa and Wotipka, 2016; Russell et al., 2018). Media can also be a source of knowledge about women’s rights (Pierotti, 2013; Swindle, 2023). Information about international precepts tends to reach people in cities because cities are more likely to contain seats of government, universities, hospitals, and therefore visitors from around the world (Hadler, 2017; Pierotti, 2013). Likewise, people may be more likely to move to cities if they have an affinity for global scripts. On the flip side, it may take longer for individuals in remote rural locations to learn of these discourses. Finally, engagement with the labor market and household wealth signifies greater incorporation into “elite” classes with broader and more cosmopolitan networks that are more likely to carry principles from the global system (Hadler and Meyer, 2009).

Another consideration relates to the content of global principles (Ulybina, 2023), and their association with different institutions. The widespread support for physical integrity rights on a national level (Cole, 2016) leads us to expect more cross-cultural consistency around rejecting violence than around women being household decision-makers. At the same time, communities that women identify with, such as religious groups, and communities where women live, provide institutional contexts that may contradict aspects of global women’s rights scripts (Thomas, 2007). Specifically, whereas global institutions promote a gender equality ideology, gender relations in Islamic and Catholic communities may take a different form, for example, stressing men as household heads (Boyle et al., 2017; Mahmood, 2005). Thus, we expect women in these communities to have a greater probability of partial rights adoption.

Finally, and taking a more micro perspective, theories of household bargaining emphasize the importance of women’s economic resources in creating gender equality. For example, outside income for women may increase their power within households by creating more deference from other household members and making women more willing to express their opinions. We expect economic resources to affect household decision-making power because many household decisions revolve around resources. When it comes to IPV, the impact of resources is not so clear. Research suggests more resources for women can actually increase IPV in some instances (Weitzman, 2014), especially when a woman’s human capital exceeds that of her partner (Svec and Andic, 2018), because this threatens the masculinity of the male partner. For these reasons, we expect women’s resources to be associated with partial-enactment privileging household decision making.

Data and methods

Data and sample

Data come from IPUMS DHS from 25 countries in Africa and South Asia (Boyle et al., 2022). The surveys are cross-sectional and ask questions of nationally representative samples of women of childbearing age (15–49 years old). The DHS uses a two-stage cluster sampling strategy to achieve representativeness. Enumeration areas (EAs) are selected based on countries’ censuses and then households are randomly selected from within each enumeration area. The DHS are typically conducted by government agencies with technical assistance from ICF International and some funding from the United States Agency for International Development. Surveys are translated into and administered in local dialects. The IPUMS version of the DHS is fully harmonized to facilitate comparative research using surveys from multiple countries and years.

We selected countries that had at least two waves of survey modules asking questions about attitudes toward IPV and decision making in the household. Survey years range from 1999 to 2017 and the average interval between waves was 9 years (the shortest was 3 years for Bangladesh and the longest was 16 for Malawi). Only women who are currently partnered are asked questions about household decision making, so we limit our sample to these women. Visitors were eliminated from the sample.

Variables

Rejection of IPV/household empowerment.

We rely on two series of questions to create the four-category empowerment variable. The first series of questions ask women if a husband is justified in “beating his wife” under five different scenarios, that is, burns food, argues with him, neglects the children, goes out without asking him, or refuses to have sex (DHS enumerators make every effort to ensure that the intimate-partner-violence module questions are asked in private, because of the questions’ sensitive nature). Based on these questions, we created a dichotomous variable coded 1 for opposition to wife beating under all circumstances and 0 if wife beating was accepted under one or more scenarios. This is consistent with previous operationalizations of the rejection of IPV (Behrman and Frye, 2021; Kurzman et al., 2019; Pierotti, 2013). A second series of questions asked women the role they play in decision making for major household purchases, for visiting family, and concerning their own personal healthcare. Women who indicated they made two or more of these decisions, either alone or jointly with their partners, were assigned a 1. Survey respondents who did not participate in decision making in at least two of these areas were assigned a 0 (for a similar approach, see Fan and Loria, 2020).¹

We combined the two resulting dichotomous variables (rejection of IPV and household decision-making power) into a four-category variable that reflects women’s orientation to global scripts: (1) dual enactment, whereby a woman rejects IPV and has a say in major household decisions; (2) partial enactment, in which a woman plays a role in major household decisions but accepts IPV; (3) partial enactment, in which a woman rejects IPV but does not play a major role in household decisions; and (4) no enactment, in which a woman neither rejects IPV nor plays a role in major household decisions (see also Table 1).

Independent variable measures.

Variables measuring exposure and amenability to global scripts include educational attainment, place of residence (urban versus rural), radio access, and household wealth. Educational attainment is measured by distinguishing between respondents with no formal education, a primary education, a secondary education, and higher education. Place of residence was coded by the interviewer based on country-specific definitions and the sampling unit designation. Radio access refers to exposure on a weekly or more frequent basis to radio. Household wealth is reported in quintiles, based on a DHS calculation that uses durable goods and assets. It distinguishes household wealth within countries rather than across countries; thus it is a relative measure (Rutstein and Johnson, 2004).

To gauge the effect of local institutions, we consider individual religious affiliations and also construct several variables that aggregate data to the enumeration-area level. The first aggregated variable is whether more than two-thirds of respondents within an EA were Muslim. Islam was chosen based on previous research at the national level suggesting that predominantly Muslim countries tend to have weaker support for gender equality (Cole, 2013; Inglehart and Norris, 2003) and an inclination toward essentialism (Mahmood, 2005). In an Africa-only model, described below, we also include an aggregate variable indicating if two-thirds of respondents were Catholic. As a measure of patriarchal norms, we aggregate the percentage of women below the age 18 when first married in each EA. For the aggregated variables, to ensure a minimum number of cases, we dropped EAs with fewer than 10 respondents. The average number of respondents across EAs was 29.

Household-bargaining-related predictors include women’s employment status, household education inequality, and age. Women who engage in work for pay outside of their own household are considered currently employed. Computing the gap between a woman’s education attainment and that of her husband or partner yields a three-category variable: a woman has less education (33%), equal education (55%), or more education (11%) than her partner. In many communities, age is associated with increased status within households, so we include this measure in our models as well.

To control for time, we operationalize two waves of surveys. The waves refer to in-country change across time; the start and end time and the time range vary depending on when surveys were administered (first wave = years 1999–2011; second wave = years 2008–2017).

Analytical strategy

We begin by exploring the general scale of women’s enactment of global women’s rights scripts: dual enactment, partial enactment, and no enactment across the 25 countries and over time. Designated weights provided by IPUMS DHS were used for all estimates, to account for the complex survey design and extrapolate to countries’ populations.

We then move to consider the effect of the numerous measures on the dependent variable. Wanting to unpack the effect of global and local institutions rather the specificity of one country over another, we opted to pool the data. We ran multinomial logistic regression analyses, comparing the relative risk ratios of each cell in Table 1 relative to the baseline category of opposing IPV and having household decision-making power (Cell 1; dual enactors). We include country fixed effects to control for unobserved country-level factors, and a dummy variable for survey-wave. We cluster standard errors at the EA level because outcomes for women within the same community may be correlated. Wald tests for combining alternatives and Hausman-McFadden test of independence of irrelevant alternatives were conducted (Long and Freese, 2014), and the results confirm the distinguishability of all the categories. We first run the model across all countries in our sample. We then divided the sample by region, to compare differences across Christians (Catholics and Protestants) in Sub-Saharan Africa.

Based on the multinomial models, we calculate average marginal effects (AMEs) of independent variables on each of the cells in Table 1. AMEs are the average of the predicted probabilities using the observed values per case (Long and Freese, 2014). To enhance interpretation, the percentage of women below the age 18 when first married and age were standardized using the following equation: x* = (x-mean)/sd. The original variables are described in Table 2, while the standardized versions are used in the analyses.

Table 2.

Descriptive statistics for the sample, weighted at the country level using population weights provided by IPUMS DHS.

	Cell 1: Dual enactment	Cell 2: Accept IPV/ Dec. maker	Cell 3: Reject IPV/ Not dec. maker	Cell 4: No enactment	N, weighted to population (in millions)	Unweighted N
Individual-level indicators
Education
None	31.2%	45.6%	41.7%	54.0%	210.0	216,840
Primary	21.6%	24.4%	21.2%	23.1%	113.7	151,690
Secondary	35.8%	26.6%	31.1%	21.4%	148.9	151,919
Higher	11.4%	3.4%	6.0%	1.5%	31.1	31,226
Listens to radio frequently	28.8%	26.1%	27.5%	30.3%	141.4	197,244
Age, mean (sd)	33.3 (8.1)	33.1 (8.2)	30.5 (8.4)	30.7 (8.4)	32.2 (8.4)	32.1 (8.4)
Religion
Muslim	33.2%	28.2%	39.5%	39.8%	172.6	176,731
Christian	21.5%	26.5%	15.0%	22.0%	109.5	210,376
Buddhist	0.6%	0.6%	0.5%	0.4%	2.7	2,934
Hindu	42.1%	42.3%	42.8%	35.3%	205.2	137,165
Traditional	0.3%	0.5%	0.4%	0.9%	2.6	7,375
Other	2.2%	1.9%	1.8%	1.6%	9.5	17,094
Woman currently working	42.2%	45.9%	38.0%	46.6%	217.8	285,981
Urban residence	40.0%	27.0%	29.3%	21.6%	154.2	184,618
Household indicators
Household wealth
Poorest	14.6%	20.3%	19.0%	24.2%	95.2	105,400
Poor	15.9%	21.3%	19.9%	23.8%	99.1	105,658
Middle	17.8%	21.7%	19.5%	22.1%	100.8	109,202
Rich	21.0%	20.6%	20.4%	18.5%	101.8	110,785
Richest	30.6%	16.1%	21.2%	11.4%	105.4	120,630
Education gap
Woman has less educ	31.0%	33.9%	34.0%	35.0%	166.6	162,941
Woman’s educ equals partner	56.0%	54.9%	54.5%	56.0%	278.4	324,897
Woman has more educ	13.0%	11.2%	11.5%	9.0%	57.2	63,837
Community-level indicators
% under 18 when married, mean (sd)	52.8 (25.3)	57.5 (23.3)	61.6 (23.1)	63.8 (21.8)	57.9 (24.1)	49.9 (23.4)
Muslim majority	29.7%	24.2%	35.5%	35.9%	153.5	159,662
Catholic majority	39.5%	30.0%	13.3%	17.2%	101.3	198,379
Sample indicator
Second wave	43.0%	36.0%	36.4%	29.3%	186	312,731
Number of Observations	34.08%	27.02%	17.90%	21.00%	502.2	551,675

Results

In both Waves 1 and 2, dual enactors (Cell 1 of Table 1) constituted the largest subgroup of women but were not the majority of women (see Figure 2). In Wave 1, 1999–2011, 31% of women were dual enactors; in Wave 2, 2008–2017, that percentage had increased to 40% of women in the weighted DHS samples. The share of women experiencing no enactment of the two global women’s rights studied here (Cell 4 in Table 1)—decreased from 24% to 17% from Wave 1 to Wave 2. In contrast, the two partial-enactment categories remained relatively stable across waves, highlighting the importance of understanding these groups. The share of partial enactors who privileged decision making (Cell 2 in Table 1) decreased by only 1.3%, standing at 26% in Wave 2. The share of partial enactors who privileged physical integrity (Cell 3 in Table 1) changed even less, decreasing by 0.5–18%. The distribution of women across the cells runs counter to our expectations that physical integrity rights would be more widely supported than women’s decision-making empowerment.

Figure 2.

Category percentages by country and wave.

Figure 2 also provides the national-level trends for all countries in our analysis. Overall, the trends seen in the pooled sample generally hold true at the country level as well. In four countries (Mali, Guinea, Senegal, and Burkina Faso), however, the share of women surveyed during Wave 1 who fully enacted global scripts—opposed IPV and were decision-makers—was very low, below 10%. On the other end of the spectrum, in two countries (Madagascar and Namibia) more than 50% of women were dual enactors at Wave 1. That said, in all but two countries (Guinea and Madagascar), the proportion of women who were dual enactors increased over time.

Multivariable analysis

Table 3 shows the results of a multinomial logistic regression with country and time fixed effects. The reference category is Cell 1 of Table 1, dual enactors. Each column shows the relative risk ratio of a woman being represented in one of the other cells in Table 1: Cell 2, privileging decision making; Cell 3, privileging physical integrity; and Cell 4, no enactment of either empowerment dimension. Figure 3 shows the AMEs for each of the four outcomes, derived from the multinomial logistic regression (Supplemental Table A1). For continuous variables, the marginal effects represent partial probability effects of a one-unit change in the independent variables. For categorical variables, the logic is similar to that of a matching study, where subjects have identical values on every independent variable except one—for that one variable, the odds compare one category against all others. In this section, we focus separately on the unique associations for each cell, drawing primarily on the AMEs.

Table 3.

Multinomial logistic regression model showing the risk of outcomes shown in Table 1 relative to outcome in Cell 1 (Dual enactment).

	Cell 2: Accept IPV/ Dec. maker	Cell 3: Reject IPV/ Not dec. maker	Cell 4: No enactment
Individual-level indicators
Primary schoola	0.91*** (0.02)	0.87*** (0.02)	0.82*** (0.02)
Secondary schoola	0.70*** (0.02)	0.78*** (0.02)	0.57*** (0.02)
Highera	0.34*** (0.01)	0.61*** (0.03)	0.21*** (0.01)
Listens to radio frequently	0.94** (0.02)	0.93** (0.02)	0.94** (0.02)
Age (standardized)	0.99 (0.01)	0.71*** (0.01)	0.73*** (0.01)
Hindub	0.90* (0.04)	1.85*** (0.10)	1.65*** (0.10)
Muslimb	0.99 (0.04)	1.75*** (0.09)	1.75*** (0.09)
Buddhistb	1.10 (0.15)	1.30* (0.17)	1.71** (0.29)
Traditionalb	1.28** (0.11)	1.25* (0.12)	1.69*** (0.16)
Otherb	0.94 (0.04)	1.65*** (0.10)	1.42*** (0.08)
Currently working	1.05* (0.02)	0.68*** (0.01)	0.78*** (0.02)
Household indicators
Urban residence	0.88*** (0.03)	0.72*** (0.02)	0.68*** (0.03)
Poorest wealth quintilec	0.99 (0.03)	0.96 (0.03)	0.97 (0.03)
Poor wealth quintilec	1.01 (0.02)	1.02 (0.03)	1.01 (0.03)
Rich wealth quintilec	0.91*** (0.02)	1.03 (0.03)	0.92** (0.03)
Richest wealth quintilec	0.63*** (0.02)	1.00 (0.03)	0.63*** (0.02)
Woman has less educd	0.90***(0.01)	1.02(0.02)	0.95* (0.02)
Woman has more educd	1.15*** (0.02)	0.97 (0.02)	1.04 (0.03)
Community-level indicators
% below age 18 when married (standardized)	1.07*** (0.02)	1.17*** (0.02)	1.24*** (0.02)
Muslim majority	1.01 (0.05)	1.22*** (0.08)	1.23** (0.08)
Sample indicator
Second wave	0.74*** (0.02)	0.72*** (0.02)	0.44*** (0.01)
Constant	0.61*** (0.04)	0.31*** (0.02)	0.23*** (0.02)
Wald chi-squared	26725.9
Observations	551675

Note: Relative Risk Ratios; Standard errors clustered at the enumeration-area level in brackets. All models include country fixed effects.

^aReference category is no education.

^bReference category is Christian.

^cReference category is middle wealth quintile.

^dReference category is woman’s education equals partner.

^*p < 0.001;

^**p < 0.01;

^***p < 0.05.

Figure 3. Average marginal effects for each of the cells, calculated from the multinomial logistic regression analysis.

Note: Horizontal lines delimit 95% confidence intervals and touching the zero line indicates insignificant coefficients. The results are calculated based on the regression analysis presented in Table 3. Percentage below age 18 when married and the two lowest household wealth categories are omitted for brevity.

The results in Figure 3 reveal that education, employment outside the home, living in a wealthy household, and urban dwelling are all consistently and positively associated with dual enactment (represented in Figure 3 with circles) and negatively associated with experiencing neither dimension of human rights (represented in Figure 3 with triangles). Women at every level of education have a higher probability of dual enactment than women with no education. Women who work outside the home are also more likely to be dual enactors; compared to women who do not work, they are 3.0 percent more likely to be dual enactors, holding all other variables constant. This is also the case for women living in households in the wealthiest or second-wealthiest quintiles. Turning to urban residency, the probability of dual enactment is 5.2 percent higher for urban women, and the probability of no enactment is 3.6 percent lower for these women. The probability of women who listen to the radio frequently being dual enactors is 1.3 percentage points higher than non-listeners, but this type of media exposure is not significantly associated with no enactment. These associations are consistent with world society theory because they suggest that amenability and exposure to global scripts links women to both the ideation and practice of women’s rights.

Other factors are negatively associated with dual enactment. Women who have more education than their partners are less likely to be dual enactors (a 1.4% decrease in probability), providing support for the idea that women must compensate for violating gender norms when their status exceeds that of their husbands. Women who identify with any religion other than Christianity are also less likely to be dual enactors and more likely to experience no enactment of either woman’s right considered here. In terms of community norms, women who live in predominantly Muslim settings are more likely to reject both dimensions of women’s rights (with a probability of 2.4 percentage points) than women in any other religious setting when individual religious orientation is controlled. Patriarchal norms—a high percentage of women married below the age of 18—shows similar results. These findings are partially consistent with claims that global scripts are based on Western value systems.

Turning next to the partial-enactment categories, the AMEs for partial enactors with household decision-making power (Cell 2 in Table 1) are represented by squares in Figure 3. Employment outside the home is associated with a 3.7 percent increase in the probability that a woman is in this category. Women who have more education than their partners are also more likely to be represented by this category (2.6% increase), while women with less education than their partners are significantly less likely (the probability is 1.7% lower). Taken together, these findings suggest that, among partial adopters of global women’s rights scripts, household bargaining power may provide women with more access to decision making but does not greatly affect their views on IPV.

A different set of variables represent local values, norms, and cultural practices that shape gender relations, potentially in ways that contrast with global scripts. The findings show that, for the category of partial enactment that privileges decision making, religious norms, and identities play a different role than is typically expected in research on women’s rights. Women who identify as Muslim or Hindu (vs other religions, including Christian), on average, have a lower probability of this form of partial enactment (5.3 and 6.9% lower, respectively). A parallel association is found for women who live in predominantly Muslim communities and communities where child marriage is more common (a decline of 1.8 and 0.5% in the probability, respectively). This means that women in these communities are less likely than women in other communities (primarily Christian in Africa and Hindu in South Asia) to endorse IPV while simultaneously holding household decision-making power.

With respect to partial enactors who oppose IPV but have little household decision-making power (Cell 3 in Table 1), college education and more household wealth increase the probability of this outcome (see Figure 3; this outcome is represented by diamonds). This may reflect a form of patriarchy in wealthier households that acknowledges the importance of treating women with respect but not necessarily empowering women within the household. In turn, women who work outside the home for pay are less likely to be represented by this category, as are women who have more education than their partners (the reduction in probability is 4.4% and 1.1%, respectively). The results also show that women who identify as Muslim or Hindu are more likely to be in this category relative to Christian women (with a probability higher by 5.1% and 6.6%, respectively), suggesting, among partial adopters of global scripts, Muslim and Hindu women are more likely to emphasize a woman’s right to be free from domestic violence. Living in a primarily Muslim community (2.0% increase in the probability) and in a community characterized by young marriages are associated with a greater likelihood of partial rights enactment privileging opposition to IPV.

To explore the role of gender relations in more depth, we repeated the multinomial logistic regression analysis separating South Asian samples and sub-Saharan African samples. Figures 4 and 5 display their AMEs, respectively; the corresponding multinomial regression tables are available in the online appendix. (Egypt was excluded from these analyses since it does not clearly fit in either region). An advantage of the separate regional models is that they allow us to distinguish between Catholic and Protestant Christians in Sub-Saharan Africa.

Figure 4. Average marginal effects for each of the cells, South Asian samples.

Note: Horizontal lines delimit 95% confidence intervals and touching the zero line indicates insignificant coefficients. The results are calculated based on a regression analysis similar to the one presented in Table 3 but using only South Asian samples. Percentage below the age 18 when married and the two lowest household wealth categories are omitted for brevity.

Figure 5. Average marginal effects for each of the cells, Sub-Saharan African samples, distinguishing between Catholic and Protestant Christians.

Note: Horizontal lines delimit 95% confidence intervals and touching the zero line indicates insignificant coefficients. The results are calculated based on a regression analysis similar to the one presented in Table 3 but using only Sub-Saharan African samples. Percentage below the age 18 when married and the two lowest household wealth categories are omitted for brevity.

Catholics were significantly more likely than Protestants to be partial enactors who privilege decision making and significantly less likely to be partial enactors who privilege bodily integrity (see Figure 5). Living in a predominantly Catholic community had an even stronger effect on the probability of women privileging decision making over opposition to IPV. Overall, this suggests Catholics are less amenable to the women’s rights script of freedom from violence and more amenable to the script of household decision-making power. Including Catholics as a separate category did not affect the associations of other religious affiliations with the outcomes. For example, Muslims were still less likely than Christian Protestants to be partial enactors who privilege decision making (see Figure 5). Individuals’ religious affiliations had stronger associations with outcomes in the South Asian model (see Figure 4) than in the African model, while dominant religion was a more potent driver of outcomes in Africa.

Discussion and conclusion

With comprehensive survey data from 25 countries, we investigate individuals’ enactment of women’s rights. While previous research has tended to treat the adoption of women’s rights as an either/or alternative, we explicitly address recent calls (Desai et al., 2022) and incorporate partial enactment of the women’s rights principles into our analysis. The SDGs recognize six dimensions of women’s empowerment; we focused on two: a woman participating in her household’s decision making and a woman’s opposition to intimate partner violence.

Across low- and middle-income countries in South Asia and Africa, our findings indicate that the percentage of women enacting both aspects of women’s rights has increased, but partial enactment—that is, enactment of one or the other principle, but not both—is more prevalent than dual adoption, even in the most recent period. Among partial enactors, enactment of household decision making is more common than enactment of opposition to IPV. This is consistent across nearly all countries.

In terms of which women are more likely to be partial adopters, local community institutions—reflecting different gender ideologies—play a role. We measured local religious norms both in terms of individual religious affiliation and majority religion within a community. The former is a reflection of a person’s identity, while the latter represents a source of external social control when individual religious identity is controlled. The associations with partial adoption are similar across the two measures. Compared to Christian women, Muslim and Hindu women are more likely to enact neither aspect of women’s rights or to be partial rights enactors privileging opposition to IPV. Women who live in predominantly Muslim communities, regardless of their own personal religious affiliation, show the same association—they are more likely to enact neither aspect of women’s rights or be partial enactors who oppose IPV but are largely uninvolved in household decision making. These results are consistent with, but add more nuance to, the understanding of how religion operates, compared to earlier studies (Charles, 2020). Our models also considered a separate measure of patriarchy—the percentage of women in a community who married before the age of 18. This indicator showed the same association with the empowerment outcomes: women who lived in such communities were more likely to be partial enactors of women’s rights, with opposition to IPV being the aspect they adopted.

This study extends previous research showing rights related to humans as embodied beings meet less institutional resistance at the national level than rights related to humans as autonomous individuals (Cole, 2016). Our findings show that partial enactors of women’s empowerment are more likely to privilege household decision making than opposition to IPV—a seeming contrast with earlier findings in this area. One interpretation is that the spread of empowerment as lived experiences operates differently than the spread of national policies when it comes to women’s rights. We look forward to future work elaborating on how and why these differences exist.

Of interest, in sub-Saharan African countries, where the DHS distinguishes Catholicism from other forms of Christianity, Catholic women reveal a distinct pattern. They, and women who live in predominantly Catholic communities, are more likely to be partial enactors of women’s human rights than Protestant women, and, unlike Muslim or Hindu partial enactors, they are more likely to endorse IPV. Thus, there is important variation in the association of different types of Christianity with partial enactment of women’s rights. This finding is noteworthy because it undercuts presumptions that Islam is the primary impediment to gender equality. In terms of progress toward eliminating IPV, resistance appears stronger from a branch of Christianity, perhaps representing a growing divide between Christian denominations that are more progressive or more conservative on women’s rights issues (Wilde, 2020). This finding also shows there are important differences among religious communities that are often grouped together as “gender essentialist” (i.e. Islam and Catholicism).

Our findings are in line with the suggestion of world society theory that exposure to and amenability toward global scripts will make it more likely women experience both dimensions of women’s rights. We find that women with more education (with one exception, noted below), who listen to the radio more frequently, live in urban areas, from a wealthy household, and work outside the home for pay are more likely to be dual enactors than other women. (Greater wealth is also positively correlated with partial adopters who privilege opposition to IPV.) These results are consistent with previous work focused on the paths of diffusion of global norms to individuals (e.g. Behrman and Frye, 2021; Charles, 2020; Hadler, 2017; Pandian, 2019). Given current trends of urbanization and the expansion of mass education, these findings may be a signal that the share of women who enact both dimensions of women’s rights will continue to increase.

An exception to the associations between world-society-related indicators and dual enactment is women who have more education than their partners. These women are less likely to be dual enactors of the two dimensions of women’s rights. One explanation is that such women are violating gender norms by having a marker of status that exceeds that of their partners. As a result, rather than gaining power from their education, they must compensate for violating gender norms by being more deferential to their partners.

Women with attributes suggesting they have greater household bargaining power (working outside the home and older), when partial rights enactors, are more likely to privilege household decision making. Household bargaining power thus appears to provide women with more access to decision making but not necessarily lead them toward opposition to IPV. The finding on the role of age reinforces earlier findings (Charles, 2020) on the association between egalitarian gender attitudes and age.

Our attention to the multiplicity of dimensions of women’s rights is also in line with Knight and Brinton (2017), who suggest gender attitudes are better described in terms of discrete multi-dimensional categories rather than a traditional-egalitarian continuum. Our findings reveal that dual enactment of globally diffused women’s rights conceptualization is on the rise. At the same time, in recent years and in tandem with the ongoing illiberal backlash (Bromley et al., 2020), women’s rights have experienced a pushback in some countries (Lerch et al., 2022). An anti-gender-equality movement is on the rise (Dietze and Roth, 2020; Korolczuk and Graff, 2018), including actors who promote the ideology that IPV is not a public issue (Bluhm et al., 2021). It is possible, if illiberal culture further diffuses worldwide, that more women will shift from dual to partial enactment of the liberal scripts of women’s rights. As noted, women’s empowerment itself is made of several dimensions, and even when focusing on two, we see that individuals often embrace one form of empowerment and not the other. Scholars from the Global South (e.g. Mahmood, 2005; Mohanty, 1984) have criticized the tendency to view gender issues from a pure Western perspective. By decentering the Western emphasis on dual enactment, we are acknowledging a broader range of gender ideologies and regimes (Chatillon et al., 2018).

Our analysis breaks ground by adding complexity to the process of adopting women’s rights, but it is not without limitations. Because of data limitations, we focus only on two regions of the world, and therefore cannot generalize our results to the world as a whole. We hope future research will extend our empirical analysis to be truly global. With respect to our empowerment measures, we cannot know if women with little household authority would like to play a greater role in decision making, or if they would see decision making as an added burden. This highlights another limitation of our analysis, that survey question responses do not necessarily reveal the meaning of social phenomenon in the local context. While it is reasonable to infer that most women would see our two women’s rights measures as measures of empowerment, we cannot be certain this would be true for all women. Finally, we focused solely on women, leaving aside men’s attitudes on the topic. Further research can examine if the illiberal pushback has a differentiating effect on men’s beliefs regarding these and additional dimensions of women’s rights.

In conclusion, this article carries general insight for world society theory. More often than not, the elements carried by and constitutive of world society—such as rationalization, scientization, and human rights—are discussed as unities. We take world society theory a step further by unpacking, if only partially, the concept of women empowerment, as presented by SDG 5. The diffusion of the empowerment script reflects the prioritization of different elements of empowerment; for example, privileging decision making or physical integrity. Acknowledging women’s agency, the notion of enactment (ranging from dual, partial, and no enactment) opens up numerous possibilities, where individuals may act upon different configurations of global culture. In sum, our approach moves away from a monolithic assumption that ties decision making and rejection of IPV together as the appropriate model of women’s rights to a more inclusive understanding. It does so by centering women who partially enact the global scripts of empowerment and investigating the power of global and localized institutions. Operating side by side, these factors contribute to varying adaptations of women rights.