In panel studies on sensitive topics, respondent-generated identification codes are often used to link records across surveys. However, usually a substantial number of cases are lost due to the codes. These losses may cause biased estimates. Using more components and linking the codes by the Levenshtein string distance function will reduce the…

Regressions can be weighted by propensity scores in order to reduce bias. However, weighting is likely to increase random error in the estimates, and to bias the estimated standard errors downward, even when selection mechanisms are well understood. Moreover, in some cases, weighting will increase the bias in estimated causal parameters. If…

A series of studies using simulation for the study of evaluation utilization is reviewed. Communication theory is used as the conceptual framework to interpret the potential influences on utilization. Implications for the practitioners are discussed. (Author/GK)

This study aims to shed light on the main characteristics of the French system for redistributing wealth to families through tax revenues and social transfers. For the purposes of this exercise, the authors used the MYRIADE microsimulation model, which covers most of the redistribution system, though it is limited to monetary flows such as family…

Studied the analytic problems associated with a design in which one identifiable group is allocated to each treatment condition and members of these groups are measured to assess the intervention. Results from a simulation study underscore the analytic problems associated with these quasi-experimental or group-randomized designs. (SLD)

Selectivity problems are discussed in terms of a general model that is estimated by the maximum likelihood method. Both single-group and multiple-group analyses are considered. An extension of the general model to latent variable models is discussed. (Author/PN)

Strategies to avoid the penalties of extra variation described by J. Cornfield and reduced degrees of freedom in community-level trials were compared in Monte Carlo simulations. The three conditions necessary to ensure nominal Type I and Type II error rates are detailed. (SLD)

A Monte Carlo study compared performance of linear and logistic mixed-model analyses of simulated community trials having specific event rates, intraclass correlations, and degrees of freedom. Results indicate that in studies with adequate denominator degrees of freedom, the researcher may use either method of analysis, with certain cautions. (SLD)

This article describes how microsimulation analysis was used to help design a social experiment in two Canadian provinces. The microsimulation was used to choose among alternative program models, to refine the selected model, and to project costs for the Canadian government's program of financial incentives for leaving welfare. (SLD)

The validity of vignettes used to elicit subject responses in social science research is examined in a study of the validity of a group of patient assault vignettes seen by 12 persons with experience in patient care. A model for establishing the validity of vignettes is proposed. (SLD)

The relationship between voting habits and parental status on the need for evaluative data before supporting a controversial school program was examined. Parents who voted most frequently wanted the most information and less immediately gave support. Nonvoting nonparents were similar. Voting nonparents were more likely to provide immediate…

Because of increased aid to developing nations, cross-cultural evaluation has become more common. Whether culture has an influence on evaluation was studied by simulating a study of Mexican and U.S. agricultural administrators. Results demonstrate the effects of cultural differences on evaluation. (SLD)