Uma análise de dados de Surveys anuais de desenvolvedores de software com ênfase em diferenças de gênero.

Abstract

The field of Information Technology (IT) has played a central role in contemporary economic and social development. However, this relevance contrasts with the persistent underrepresentation of women and with structural inequalities that affect their participation and recognition in the sector. In light of this scenario, this research focused on investigating gender disparities among IT professionals, considering attributes such as experience, salary, and geographic location. The analysis was based on data from Stack Overflow’s annual survey and employed a quantitative approach divided into two complementary studies. The first study focused on the relationship between gender and job role, applying hypothesis testing, confidence intervals, and regression models to identify salary disparities associated with specific roles, as well as conducting a temporal analysis of the results. The second study incorporated variables such as professional experience and geographic location, using the Association Rule Learning (ARL) technique combined with hypothesis testing to identify recurring association patterns between gender and professional profile characteristics. The results reveal consistent salary inequalities that disadvantage women, even among professionals with similar profiles. The association rules also highlighted systematic correlations between the female gender and lower salary ranges, especially in certain regions of the world. The combination of robust statistical methods strengthened the reliability of the analyses. The findings of this research contribute to advancing academic discussions on gender equity in technology and offer practical insights for organizations committed to creating more inclusive environments. Finally, the study’s limitations are discussed, and directions for future research are suggested, such as expanding the set of analyzed variables, incorporating qualitative methods, and replicating the approach using other datasets.