Multiple imputation of categorical data

19-7-00

Click to start

Contents Multiple imputation of categorical data TNO applications of MI Overview Imputation of categorical data MICE Gibbs sampler Imputation by MICE Imputation with categorical data Some models for categorical data Proportional odds model Generalized logit model Estimates Four steps to impute categorical data Some data... Ppt-dia Missing data pattern Missing data by age group (PH) Analysis 1: Will MI ‘repair’ the data? PH stages 5 and 6: Age distribution PH stages 5 and 6: Counts Analysis 2: Will inference be OK? Generalized logit: Coverage 95%c.i. Conclusion Software

Authors: K. Oudshoorn, S. van Buuren & J.L.A. van Rijckevorsel E-mail: K.Oudshoorn@pg.tno.nl;S.vanBuuren@pg.tno.nl