Introduction to Special Section: Quantitative Approaches to the Measurement and Analysis of Female Empowerment and Agency

Abstract

This special issue presents four novel applications of quantitative methods to address measurement and analytical issues in the appraisal of female empowerment and agency. The methods presented comprise mixed methods, dominance analysis and structural equation models. The use of these methods is illustrated with empirical applications in Cambodia, India, Mali, and Turkey.

Public commentators are not tiring of telling us that we live in interesting times. Whether it is political upheaval in the Western World or war in the Middle East, we live in interesting times. In the midst of these fast-pacing events, it is important not to overlook a very positive trend: collective action, changes in attitude, and policies favouring an expansion of female agency and freedom. From relaxation of China’s ‘one-child’ population policy, to ditching the ban against women driving in Saudi Arabia, to a healthy increase in disapproval of sexual harassment in high places, to the hardening of laws against femicide worldwide, especially in Latin America, it seems that large swathes of the world are heeding the call of the United Nation’s fifths Sustainable Development Goal to ‘End all forms of discrimination against all women and girls […] Eliminate all forms of violence […] Eliminate all harmful practices […]” (UNDP, 2017).¹

From a capability approach perspective, achieving gender equality is inconceivable without expanding women’s agency. The definitions of female agency and autonomy revolve around women’s capability to influence their fate and surroundings enjoying both positive and negative freedoms (Kabeer, 2005), and including control of resources, setting their own goals and acting upon them (Kabeer, 1999; Malhotra & Schuler, 2005). A related word, empowerment rather denotes the improvement process from lack of agency to agency (see Ballon, 2017, in this special issue and references therein).

Given the many realms of life relevant to people – health, personal relationships, finances – female agency is necessarily a multifaceted phenomenon (Narayan, 2005). This has several behavioural and policy implications. For instance, in some societies women may be empowered in some dimensions but not in others (Malhotra & Schuler, 2005), while women may place different importance across dimensions (Kabeer, 1999). Likewise, the multidimensionality of female agency poses measurement and analytical challenges, some of which are addressed in this special issue.

The academic literature has also long acknowledged both the intrinsic and instrumental roles of female agency and empowerment implicit in the sustainable development agenda. Intrinsically, female empowerment is justified both negatively, in the form of ending abuse and discrimination, and positively, as the expansion of capabilities to live the lives women have reason to value (Eswaran, 2014). Perhaps more importantly, people may value agency for its own sake, such as wielding control over the decisions affecting their lives, although empowerment measures may not always correlate strongly with improvements in women’s life satisfaction, which could partly be due to well-known adaptation effects (Eswaran, 2014). Similarly, several instrumental roles of female autonomy are well established in the literature, including fostering entrepreneurial activity, family planning, health and education in the family, and so forth (for example, see Duflo, 2012, and references therein).

In this special issue, we present four novel applications of quantitative methods to address measurement and analytical issues in the appraisal of female empowerment and agency. The methods here comprise mixed methods (Klein & Ballon, 2017), dominance analysis (Chaudhuri & Yalonetzky, 2017) and structural equation models (Ballon, 2017; Yilmaz, 2017). Mixed methods have shown to offer a more holistic view of measurement of female empowerment and agency (c.f. Rao & Woolcock, 2005). Stochastic dominance techniques were originally applied to financial decision-making under risk (for example, Levy, 2015), but soon found manifold applications in the fields of social welfare comparisons and distributional analysis. Since the pioneering work of Atkinson and Bourguignon (1982) stochastic dominance techniques have been extended to the domains of joint distributions with ordinal data in applications to welfare comparisons (for example, Yalonetzky, 2013), inequality comparisons (for example, Decancq, Fleurbaey, & Schokkaert, 2014), and poverty comparisons (for example, Permanyer & Hussain, 2017; Yalonetzky, 2014). Structural equation models have been widely applied in psychometric studies (for example, MacCallum & Austin, 2000), marketing research (for example, Martinez-Lopez, Gazquez-Abad, & Sousa, 2013), civil engineering (for example, Xiong, Skitmore, & Xia, 2015), and in development economics as well (for example, Di Tomasso, 2007; Krishakumar & Ballon, 2008; Kuklys, 2005).

In what follows we briefly summarise the papers of this special issue, with an emphasis on the method the paper uses to address measurement and analytical challenges in the quantification of female agency and empowerment.

In ‘Rethinking Measures of Psychological Agency: A Study on the Urban Fringe of Bamako’, Elise Klein and Paola Ballon (2017) examine the use of theoretical measures of psychological agency against local concepts of psychological agency. Focusing on women in the urban fringe of Bamako, Mali, the authors revisit the notion of psychological agency, and internal motivation and question the validity of Western theoretical concepts of psychological agency. Specifically, the authors compare components of the renowned Relative Autonomy Index in the domains of household decision-making and health against self-reported levels of two local notions of agency: dusu (loosely translated as internal motivation) and ka da I vere la (loosely translated as self-belief).

Remarkably, the authors find no clear positive association between autonomy as measured by the Relative Autonomy Index and reported levels of the two local notions of agency. The results are robust to alternative weighting schemes for the construction of the index. By contrast, the authors do find strong and statistically significant positive associations between local measures of agency (that is family-related dusu, community-related dusu, family-related ka da I vere la, and community-related ka da I vere la). The authors convincingly conclude that ‘the theoretical and local measures of autonomy are different concepts and thus measure different aspects of psychological agency. Therefore, any measurement exercise of psychological agency requires a comprehensive approach that includes local measures of autonomy together with theoretical ones’.

In ‘A Structural Equation Model of Female Empowerment’, Paola Ballon (2017) proposes a structural equation model to measure and explain female empowerment. Building on the capability approach and in the intra-household gender dynamics literature, the author proposes a more holistic and comprehensive conceptual framework for quantifying female empowerment. As such, empowerment is defined as the decision-making ability of a woman regarding her strategic and non-strategic life choices where resources, values/traditions, and decision-outcomes are the key elements characterising a woman’s ability to choose. Following Kabeer (1999), the author distinguishes between resources, agency, and decision-outcomes. Whereas agency is defined as a woman’s ability to define her goals and act upon (or toward) them, decision-outcomes are themselves manifestations of this scope or ability to choose, and resources are seen as pre-conditions enhancing the ability to choose. Considering that empirically both resources and decision-outcomes are observable but agency is not, Ballon specifies a structural equation model (SEM) to measure and explain female empowerment in Cambodia using its 2005 Demographic and Health Survey. Indeed, SEM is particularly well suited to situations in which the phenomenon of interest, for example, female empowerment, can be deemed a latent variable only observable indirectly through a set of indicators. In the model, female empowerment poses as a latent variable, imperfectly mirrored in the observed decision-outcomes; shaped by ‘resources’, and determined by values and traditions, all modelled as exogenous factors.

The empirical application to Cambodia relies on indicators for three realms of life: self-strategic choices (for example, spousal selection), familial strategic choices (for example, fertility decisions), and second-order economic choices (for example, labour force participation). The results of the model are quite interesting and informative for public policy purposes. The main finding indicates that, educating the father largely improves a woman´s agency, while living with in-laws negatively affects her agency over her children’s health. Moreover, positive attitudes toward wife-beating restrict women’s actual freedom of movement. To provide an overall assessment of empowerment of females in Cambodia across the three realms of life considered in the analysis, Ballon exploits the model’s in-built ability to predict a score of female empowerment for each life domain. She then uses these scores to compare their distributions through first-order stochastic dominance analysis. Remarkably, the author finds that, in Cambodia, women are more empowered in the realm of self-strategic choices vis-à-vis familial-strategic ones.

In ‘Female Autonomy, Social Norms and Intimate Partner Violence Against Women in Turkey’, Yilmaz (2017) proves further the usefulness of structural equation modelling in the study of the two-way relationship between intimate partner violence and female autonomy. Relying on the 2006 Turkish Household Structure Survey, the author finds a significant and negative effect of violence against female spouses on their level of autonomy, measured by participation in household-related decisions. Simultaneously, the incidence of intimate partner violence also diminishes with higher female autonomy.

Interestingly, Yilmaz also analyses the impact of social norms on intimate partner violence. Specifically, the author finds that intimate partner violence is an increasing function of the husband’s commitment to social norms, upholding traditional gender roles characterised by women’s subordination and restricted agency. Such operationalisation of the interaction between female autonomy, violence, and social norms, renders this study the most extensive analysis of women’s empowerment in Turkey to date.

In ‘The State of Female Autonomy in India: A Stochastic Dominance Approach’, Chaudhuri and Yalonetzky (2017) tackle a seemingly straightforward, but rather tricky, descriptive question: do Indian states and regions differ in their observed measures of female autonomy? The authors address two key methodological challenges in their attempted answer: (1) the multidimensional nature of the concept; and (2) its frequent measurement with ordinal variables. The first challenge means that in practice there will be several indicators capturing relevant aspects of female autonomy, begging the question as to whether and how to aggregate them in order to produce a picture of individual female autonomy (in turn necessary for social measures). The second challenge implies that the indicators are not really amenable to direct quantitative comparisons of social averages as long as the scales attributed to the ordinal categories remain arbitrary. Yet the authors show that social comparisons are still possible resorting to stochastic dominance techniques suited for multiple ordinal and dichotomous variables. Whenever these dominance conditions hold for a pairwise comparison, we can conclude that the multidimensional autonomy distribution in one Indian state is more desirable than in another one in terms of a broad range of criteria for the individual and social welfare evaluation of autonomy.

The authors propose and apply three stochastic dominance methods, each reflecting alternative yet meaningful welfare comparison criteria, to India’s National Family Health Survey 2005–2006. The dataset contains questions on women’s say (or lack thereof) over several domains of their lives, ranging from family visits to large household purchases. Consistently across the three methods, they find that most of the states with better autonomy distributions (in pairwise comparisons) come from the north east (that is eastward beyond the Siliguri corridor) and the south, whereas most of the states with worse autonomy come from the north. Though the results corroborate previous findings emphasising better female autonomy among less ‘sanscritised’ states, the authors do find several interesting exceptions in state-to-state comparisons. Hence it is not always the case that any random southern or north-eastern state will ‘dominate’ a northern state (that is, exhibit better female autonomy distribution). Moreover, and less documented in the literature, they find that generally (albeit with exceptions) north-eastern states feature better female autonomy distributions than southern states.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. http://www.undp.org/content/undp/en/home/sustainable-development-goals/goal-5-gender-equality.html.