Cover Image
QR Code

Imputing Missing Data: A Comparison of Methods for Social Work Researchers.

A study compared six methods of data imputation used to handle the problem of missing research data: listwise deletion; mean substitution; hotdecking; regression imputation, sometimes referred to as conditional mean imputation; and single implicate and multiple implicate data sets. Results suggest...

Full description

Published in:Social Work Research Vol. 30; no. 1; pp. 19 - 32
Main Authors: Saunders, Jeanne A., Morrow-Howell, Nancy, Spitznagel, Edward
Format: Article
Published: National Association of Social Workers March 2006
Subjects:
Social services > Research
Missing observations (Statistics)
Statistics
Social services
Methodology
Online Access:View this record in EBSCOhost
Summary:A study compared six methods of data imputation used to handle the problem of missing research data: listwise deletion; mean substitution; hotdecking; regression imputation, sometimes referred to as conditional mean imputation; and single implicate and multiple implicate data sets. Results suggest that single and multiple implicate methods produce more accurate values than the other methods. The methods are compared and recommendations for dealing with missing data are presented.