Investigating the Impact of Response-Style Heterogeneity on the Measurement of Psychological Constructs.

다국어 초록 (Multilingual Abstract)

The frequent use of self-report rating scale surveys in measuring psychological attributes makes the validity of this form of assessment an important consideration in social science research. One concern with this form of assessment is the possibility that respondents use the rating scale differently in responding to the items. Commonly, respondents display tendencies to over- or under-utilize rating-scale categories in a systematic way, tendencies referred to as response styles. A common example of a response style is one in which the respondent over-selects the extreme rating scale endpoints. When response-style heterogeneity is present and not accounted for, there is a potential for significantly biased estimates of the intended-to-be-measured psychological trait. In this dissertation, I consider two applications related to response-style interference with measurement of the psychological trait. The first application introduces the possibility that response-style heterogeneity at the individual respondent level as a source of person misfit. In this application I introduce a procedure for determining whether a given respondent is demonstrating a response style, as well as a model for clarifying the type of response style as well as the nature and amount of response-style heterogeneity in a population of respondents. The second application attempts to better understand the biasing implications of response styles on the estimated correlations between psychological traits. In this application, I extend findings from a previous study to demonstrate how certain test conditions can make the biasing effects of response style much greater than previously thought.