A Kernel Density Analysis of Continuous Typical-Response Scales

Abstract

The present study assessed how respondents use continuous scales in two practical measurement situations. Responses to the continuous scales were expected to be multimodal, based on the assumptions that (a) respondents do not take full advantage of the scale and (b) the selective process is relatively general and common to all respondents. Kernel density analysis and a factor-analytic procedure were used to evaluate the assumptions in two empirical datasets. In both cases, the results suggested that respondents were operating with a limited number of response points in spite of the continuous nature of the scale.