Considerations for running and interpreting a binary logistic regression analysis – a research note

Abstract

This research note discusses key considerations for analysis of categorical data using a Pearson’s chi-square and binary logistic regression. It draws on experience from analysis of a country-level household survey (Northern Ireland Health Survey 2014/15), using SPSSv25, that examined the relationship between household food insecurity status and identified demographic predictors, using Pearson’s Chi-Square test to check associations and binary logistic regressions to derive the predictive models. This note presents an overview of the assumptions for both tests which must be satisfied to ensure the tests are appropriate, discusses the usefulness of using Pearson’s Chi-Square test as a preliminary test before using binary logistic regression, and presents an overview of how to interpret the output from a binary logistic regression model.