Gender differences in decision making is a topic that has attracted much attention in the literature and the debate seems to be inconclusive. In a recent study, Bouchouicha et al. (2019) using data from an incentivised experiment with almost 3000 students and 30 different countries, estimate gender effects assuming four commonly employed definitions of loss aversion. Despite the fact that their analysis is based on the same data and the same functional forms and econometric setup, their results are inconclusive regarding the existence and the direction of gender effects for loss aversion. In this study, we investigate two extensions of their work in an effort to shed some light on the potential reasons behind this contradictory result. In particular, we explore whether: (1) a more flexible estimation method that allows for individual heterogeneity and generates more robust estimates in the presence of noise and; (2) a different utility function, can generate more robust inference regarding gender effects. We show that while a more flexible Hierarchical Bayesian estimation method is not sufficient to explain the contradictory results, an alternative utility function detects a uniform gender effect, with women being always more loss-averse, regardless the adopted definition of loss aversion.