Journal of Official Statistics, Vol.19, No.2, 2003. pp. 153–176
Prevention and Treatment of Item Nonresponse
Edith D. de Leeuw, Joop Hox, and Mark Huisman
Abstract:Item nonresponse is a problem that frequently occurs in survey data. Although it can never be prevented totally, it can be reduced considerably. This not only provides researchers with more data to use in their analyses, but also with helpful auxiliary information for better imputation and adjustment. Understanding, prevention, and imputation are chained together in handling missing data successfully. To supply survey researchers with adequate tools we will review all three aspects. First, we present a typology of missing data patterns and their origins. Based on this typology and the potential sources of item nonresponse, we outline how missing data can be prevented as much as possible. Finally, we discuss how knowledge of the data collection process can improve the statistical treatment of the remaining missing data.
Keywords:Causes of missingness; data collection mode; ignorability; imputation; item non response; questionnaire development; follow-up survey.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group