Unit of Analysis

Initially, the full IFLS-5 dataset includes over 36.391 individuals. However, to ensure sufficient statistical power for detecting the effect of household decision-making autonomy on depressive symptoms, a Monte Carlo simulation was conducted. The simulation modeled a logistic regression with autonomy as the independent variable and depression status as the binary outcome. Based on 500 replications per sample size, the simulation showed that a sample size of N = 200 yielded an estimated power of 0.83, exceeding the conventional threshold of 0.80 (Cohen, 1992). This suggests that the available sample is sufficiently powered to detect a moderate effect, assuming appropriate model specification.