Re-analysis using Inverse Probability Weighting and Multiple Imputation of Data from the Southampton Women’s Survey

In this document I describe a re-analysis of data from the Southampton Women’s Survey (SWS). The original complete-case analysis implicitly assumed that the data are missing completely at random. Inverse-probability weighting (IPW) and multiple imputation (MI) are more sophisticated methods for handling missing data, which make the weaker assumption that the data are missing at random. We sought to determine whether these methods changed the conclusions of the original analysis. We describe IPW and MI, give the results of applying them to the SWS and, where the estimates differ from the original estimates, provide intuition as to why these differences arise. We find that, assuming the missing data are missing at random, the conclusions of the original complete-case analysis remain unchanged. Thus, confidence in these original conclusions is strengthened. We hope this document may prove helpful to researchers wishing to use IPW or MI on their own data.