No abstract

No abstract

Vol. 88, No. 3 reanalyses earlier studies on political participation, vote choice, candidate evaluation, and policy position, incorporating correction for nonresponse. He finds that the coefficients for most of the models change only slightly, leaving measures of relative importance approximately the same. But coefficients sometimes double upon correction, making for far different substantive conclusions, especially on issues of turnout and information. I find this section fascinating, but am somewhat concerned with the uncritical acceptance, without any comment, of the restrictive and sometimes dubious assumptions necessary for the selection models.Finally, Brehm simulates what would happen if response rates were to drop considerably from their current levels by comparing results from NES as he successively drops out those who were the most difficult to contact or to convince to participate. His major finding is that if response rates should fall below about 30%, nothing, not even the modeling of nonresponse, could rescue the results of surveys. He closes with the plea to survey organizations to make their surveys interesting to respondents and to collect and disseminate the collateral information on respondents and nonrespondents that would make modeling of selection possible.Brehm's work is careful and comprehensive. Hence, the lack of reference to Incomplete Data in Sample Surveys (1983), the three-volume report of the Panel on Incomplete Data of the Committee on National Statistics of the National Research Council surprised me. Brehm's volume provides a great deal of important analysis for survey researchers and political scientists to think about and to act upon. and hierarchical collectivism. The following table charts his fourfold typology:

No abstract

No abstract