Measuring women’s empowerment is getting more attention than ever, and spurring debate. One challenge for researchers is to identify items that are validated across settings and over time so the impacts of programs for women’s empowerment can be tested and compared. This blog post points to some lessons we [1] learned when faced with the challenge to validate measures of women’s empowerment across contexts and over time. We are exploring these issues empirically with data from the Gender Agriculture and Assets Phase 2 (GAAP2) program team at the International Food Policy Research Institute (IFPRI).
In a study funded by the National Institutes of Health (NIH), we looked at how major events in a woman’s life are related to what we called her long-term “economic empowerment”. We investigated the Women’s Agency Scale (WAS) using 2006-2012 panel data for 4,012 married women from the Egyptian Labor Market Panel Survey. We were interested in women’s age at first marriage, and we wanted to see whether women who married later in life had higher long term economic empowerment.
This question posed a few methodological challenges. First, we had to define “economic agency” in a meaningful and measurable way. Our definition encompassed women’s engagement in the work force, and participation in major economic decisions in the household. We then had to operationalize these constructs and validate the measures in national level longitudinal surveys administered in 2006 and 2012. We then had to validate these measures longitudinally. These steps were critical, to ensure that women were interpreting and responding to questions in the survey in the same way over time. This comparability would help us interpret change in women’s economic agency over time.
A critical lesson here is that our data collection tools and methods must capture the same concept over time, or display longitudinal measurement invariance, if we want to monitor changes in women’s economic agency – or empowerment – without ambiguities.
Monitoring women’s empowerment over time requires us to create and use measures that are as free as possible of identifiable response biases. Why is this important? Some monitoring systems repeatedly expose women to the same instruments to measure their empowerment. This exposure could change the way women understand, interpret, and respond to questions. We want to avoid questions that may induce such change.
Moreover, women’s attitudes, views of the world, knowledge, as well as their understanding of their rights change over time, and these changes may influence the way they respond to questions about their own empowerment. We, as researchers, must be careful to design and to test instruments with this possibility in mind. For example, the more women become aware of their circumstances, the more they may be willing to report certain experiences, such as violence. Similarly, their awareness of their own rights (e.g. owning land or inheritance) may change. Women might also internalize harmful community norms about their place in that society and in the world, and might fear voicing views that contradict these norms. So how can we, as researchers, isolate these “response effects” that may differ across settings and over time from the impacts of programs designed to empower women?
Another example of the challenges is useful. A common way of measuring a woman’s agency is to ask ”who has the final say” in decisions about specific activities. We routinely assume that all women understand and interpret these questions similarly, but these questions are subject to some challenges. E.g.
As researchers, we need to consider these complexities when designing and using state-of-the-art tools to measure women’s empowerment consistently across settings and over time.
To measure women’s empowerment in ways that allow meaningful comparison across settings and over time, we must tease out potential response biases that could influence our comparisons across groups and our interpretations of trends. Applying state-of-the-art methods in data collection and analysis to address these complexities is important if we want to understand the impact of development programs on women’s empowerment. Validating measurements of women’s empowerment over time and across contexts is challenging, but not impossible.
[1] Kathryn Yount and research team at Emory University http://gaap.ifpri.info/gaap2-team/external-advisory-committee-2/
Kathryn Yount is Professor and Asa Griggs Candler Chair of Global Health at Emory University. This blog post was written for the EnGendering Data and Methods Blog managed by the CGIAR Collaborative Platform for Gender Research (2017-2019) and initially posted on www.gender.cgiar.org.
