Showing papers on "Imputation (statistics) published in 1982"

Models for Nonresponse in Sample Surveys

Abstract: The literature on the analysis of incomplete data using models is reviewed in the context of nonresponse in sample surveys. The modeling approach provides a large body of methods for handling unit and item nonresponse, some of which cannot be derived from the randomization theory of inference for surveys. Key concepts from the literature on incomplete data, such as factorizations of the likelihood for special data patterns, the EM algorithm for general data patterns, and ignorability of the response mechanism, are discussed within the survey context. Model-based procedures are related to common methods for handling nonresponse in surveys, such as weighting or imputation of means within subclasses of the population.

Imputation of Missing Values When the Probability of Response Depends on the Variable Being Imputed

Abstract: A method is developed for imputing missing values when the probability of response depends upon the variable being imputed. The missing data problem is viewed as one of parameter estimation in a regression model with stochastic censoring of the dependent variable. The prediction approach to imputation is used to solve this estimation problem. Wages and salaries are imputed to non-respondents in the Current Population Survey and the results are compared to the nonrespondents' IRS wage and salary data. The stochastic censoring approach gives improved results relative to a prediction approach that ignores the response mechanism.

Imputation in Surveys: Coping with Reality

Abstract: In surveys a response may be incomplete or some items may be inconsistent or, as in the case of two-phase sampling, items may be unavailable. In these cases it may be expedient to impute values for the missing items. This article discusses the general problem of non-response and the problem of imputation. Methods of imputation are reviewed, and the evaluation of the effects of imputation on the survey estimates and the monitoring of the imputation process are discussed.

An analysis of alternative imputation strategies for individuals with partial data in the National Medical Care Expenditure Survey.

Abstract: Data collection in the National Medical Care Expenditure Survey was applied to the same panel of sample households in six rounds of interviewing, with 1977 as the reference period. Approximately 11 percent of all survey participants provided data for only part of the time they were eligible to respond. To allow for national estimates of relevant health parameters, the data for the partial participants must be adjusted for the entire time frame for which they were eligible. Consequently, three alternative imputation strategies were considered for implementation: a weighted adjustment to the partial data, a substitution of data from complete participants who matched the partial respondents on relevant demographic characteristics, and use of only the data from participants with complete information to characterize the nation. To determine the optimal strategy, a controlled experiment was conducted by artificially creating partial data for participants with complete information, then adjusting the synthetically produced partial data by the three imputation strategies.

Imputing Income in the CPS: Comments on 'What Do We Know about Wages: The Importance of Non-Reporting and Census Imputation' by Lillard, Smith and Welch.

Abstract: : Nonreporting of income in the Current Population Survey is an important problem affecting the many researchers using the data base. This paper discusses an approach to handling this problem proposed by Lillard, Smith and Welch, which applies selection models to a Box-Cox transformation of the income variable. Topics considered here include: the inadequacy of single imputation and the desirability of multiple imputation, the importance of the distinction between ignorable and nonignorable nonresponse, the sensitivity of inference to assumptions unassailable by the data at hand, and the possibility of using the CPS-SSA-IRS Exact Match File to study such assumptions. The Lillard, Smith and Welch paper accompanied by this discussion is to appear in a book presenting the proceedings of the NBER Labor Cost Conference to be published by the University of Chicago Press. (Author)