The Social Geography of Women’s Attitudes toward Wife-beating in Ethiopia: A Contribution Towards Proper Application of Spatial Statistics

Spatial statistical measures have been applied to Ethiopia’s Demographic and Health Survey data (EDHS), mostly at the national level. However, there is concern that most applications violate basic principles of statistics regarding autocorrelation, or are not cognizant first law geography which states all things related but near more related. This study investigates local variations in attitudes toward wife-beating Ethiopia with education as main correlate. It does so by using a spatial measure known geographically weighted regression (GWR) appropriate conditions geographic non-stationarity than ordinary least squares (OLS). Equally importantly, it examines appropriateness existing OLS-based studies EDHS data. We found inappropriately OLS despite findings spatially autocorrelated The GWR model showed an association between acceptance educational status. also generated list twelve sampling clusters where women respondents stated was acceptable while admitting having had no formal education, R2s exceeded 0.5 modeling involving 72 nearest neighbors per cluster. An education-focused bi-variate rather multi-variate avoided issues multicollinearity keeping simple its results actionable. Although majority Harari Wereda Kilil, got their name from members ethnic group predominantly Muslim, difficult pinpoint factor set factors can be cited causally associated characteristics placed them on list. makes methodological contributions sociodemographic populations, especially those developing countries such show significant variations. adds literature regression.