On the malleability of gender attitudes: Evidence from implicit and explicit measures in Tunisia

In many regions of the world, significant parts of society are persistently unsupportive of female empowerment. The role of women is often still defined by social norms, rather than legal rights, hampering economic development. Women's empowerment has therefore become a top priority on development agendas, also testified by an increasing number of policy interventions aiming to promote gender equality. To monitor progress in this area we need reliable data on gender attitudes. However, standard self-reported measures of gender attitudes are prone to a wide range of measurement errors and social desirability bias. In this paper we address this problem and use a new field application of the implicit association test (IAT), next to a set of standard survey questions, to measure implicit gender attitudes in Tunisia. Implicit attitudes are considered less susceptible to measurement bias and may serve to more accurately assess gender attitudes. Further, we examine the malleability of implicit gender attitudes using a randomized video intervention illustrating real-life gender reforms in Tunisia, and natural variation in interviewer characteristics with respect to gender and perceived religiosity. Our study finds that the video has no average impact on implicit (IAT-based) attitudes, which is consistent with the idea that in a highly polarized society like Tunisia such an intervention only affects specific groups in a society. We indeed find that the video mitigates the implicit gender bias only among the specific subpopulation of conservative women. We also confirm the presence of interviewer effects. Yet, impacts are more pronounced for explicit attitudes, which may suggest social desirability at work in surveys. We believe that our study may inform policymakers on the potential power of light interventions and helps improve measurements related to gender norms and attitudes. (C) 2020 The Authors. Published by Elsevier Ltd.