Understanding Public Opinion in Post-Communist States: The Effects of Statistical Assumptions on Substantive Results

Examines the effects of common assumptions about causality on the substantive results of quantitative survey data analysis in studies of postcommunist mass attitudes. It is noted that most of this research depends upon a single-stage causal model that uses OLS linear regression estimation without giving adequate attention to measurement errors or other problems associated with these models. Logical, interpretation, & statistical difficulties involved in single-equation approaches to evaluating public opinion are discussed. A 1998 study of support for the government on the part of ethnic minorities in Kazakhstan is described. Data obtained from a survey of 798 Russian-speaking members of ethnic minorities were analyzed using a structural equation model (SEM) & the Amos statistical program for SEM estimations. A comparison of the results to those from a single-equation statistical approach shows that different models of causality with the same variables can generate very different outcomes. It is argued that SEM is a better approach for accurately assessing causality between demographic & attitudinal variables normally examined in postcommunist studies. 2 Tables, 2 Figures. J. Lindroth