ERIC Number: ED238948
Record Type: RIE
Publication Date: 1983-Aug-17
Reference Count: 0
The Effectiveness of Hot-deck Procedures in Small Samples.
A simulation study was conducted to identify the best hot-deck variation to impute missing values. The three variations included in the study were the hot-deck random, the hot-deck sequential, and the hot-deck distance. The properties of these methods were investigated under three levels of the proportion of incomplete records and four levels representing the number of missing values in a record. The sample size and the covariance in the samples generated remained constant throughout the study. Each cell of the design matrix was replicated 500 times. The three hot-deck variations were found unbiased estimators of population means. Compared to two other variations, the hot-deck distance method performed better in retaining the population covariance structure in imputed samples. In terms of the quality of missing value estimates, the hot-deck distance seemed better than the hot-deck sequential and the hot-deck random methods. The results of this study were found useful in choosing the appropriate hot-deck variation for imputation. (Author)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: Data Sets; Hot Deck Procedures; Missing Data; Robustness
Note: Presented at the Annual Meeting of the American Statistical Association (Toronto, Canada, August 17, 1983).