Inspiration
We were inspired to explore how social norms influence real-world outcomes. Domestic violence is widely studied in terms of incidence, but we wanted to understand the attitudes behind it. Since norms often shape behavior and policy, we focused on identifying which demographic factors are associated with higher acceptance of domestic violence across countries.
What it does
Our project analyzes Demographic and Health Survey data to examine how education, employment, residence, age, and marital status relate to acceptance of domestic violence. Using exploratory analysis and regression modeling, we identify which demographic groups show higher justification rates and quantify how much variation demographics explain overall.
How we built it
We began by cleaning and structuring the dataset, ensuring categorical variables were properly encoded. We performed exploratory data analysis to examine distributions and group differences. Then, we implemented a multiple linear regression model and conducted ANOVA to evaluate which predictors significantly explain variation in acceptance levels. We assessed model performance using R², RMSE, and MAE to understand predictive strength.
Challenges we ran into
Data preparation was one of the biggest challenges, especially ensuring consistent variable naming and encoding. Interpreting statistical output correctly (particularly distinguishing between statistical significance and effect size) required careful analysis. We also had to avoid misleading conclusions from raw counts versus percentages and ensure our interpretations stayed within association rather than causation.
Accomplishments that we're proud of
We built a model that explains over half of the variation in acceptance levels, which is strong for social science research. We successfully identified education as the strongest demographic predictor and supported that conclusion with both exploratory and statistical evidence. We also developed a clear narrative connecting data insights to policy relevance.
What we learned
We gained deeper experience in regression modeling, ANOVA interpretation, and model evaluation metrics. More importantly, we learned how to communicate statistical findings clearly and responsibly, especially when working with sensitive social issues. The project reinforced the importance of effect size, model assumptions, and careful data storytelling.
What's next for Acceptance of Domestic Violence
Future work could incorporate country-level variables such as GDP, gender inequality indices, or policy indicators to capture broader structural influences. We could also explore multilevel modeling to better account for cross-country variation. Expanding the dataset across additional years would allow us to examine how attitudes change over time and assess the impact of educational or policy interventions.
Log in or sign up for Devpost to join the conversation.