Alternative Measures of Political Tolerance: Must Tolerance be "Least-Liked"?

Abstract

In this article I consider how to measure political intolerance. The traditional Stouffer-based measure of intolerance is compared to the Sullivan, Piereson, and Marcus least-liked measure using data from a national survey. While the alternative measures of intolerance are not strongly intercorrelated, apparently the lack of correlation is largely a function of random measurement error. This conclusion is based on the empirical finding that the traditional predictors of intolerance perform very similarly irrespective of which of the tolerance measures is used. Since substantive conclusions about the origins of intolerance are insensitive to the index employed, I argue that future tolerance research can profitably utilize either measurement approach. Roughly a decade ago, the study of political intolerance within the mass public was changed rather dramatically. This change came about as a result of a new measurement strategy devised by Sullivan, Piereson, and Marcus (1979, 1982). They argued that the conventional measures of political intolerance-first constructed by Stouffer (1955)-were no longer useful. Driven by a conceptualization of intolerance as requiring an objective precondition, they developed a content-controlled set of indicators of the concept. This approach allows the survey respondents to select for themselves the group that is the target of intolerance. This new approach to political intolerance is not just an issue of refining measurement: based on this measure, basic debates about how much political tolerance exists and how it has changed have erupted. Moreover, many of the traditional predictors of intolerance have been found to be of little utility. Thus, the measurement issue has restructured substantive theory. When such an important change in how we operationalize a basic concept like political intolerance occurs, it is reasonable to expect a fair amount of technical work to validate the new measurement strategy, assess its reliability, determine how the new and Qld measures are connected, and so on. Yet with just a