Multiple imputation for unit-nonresponse versus weighting including a comparison with a nonresponse follow-up study

Abstract

"The results of a national fear of crime survey are compared with results following the use of different nonresponse correction procedures. We compared naive estimates, weighted estimates, estimates after a thorough nonresponse follow-up and estimates after multiple imputation. A strong similarity between the MI and the follow-up-estimates was found. This suggests, that if the assumptions of MAR hold, carefully selected and collected additional data applied in a MI could yield similar estimates to a nonresponse follow-up at a much lower price and respondent burden." (author's abstract)