Decomposing Variance Components for Risk Perceptions Using Generalizability Theory
Date of Award
Doctor of Philosophy (Ph.D.)
Scott Highhouse (Advisor)
Margaret Brooks (Committee Member)
Richard Anderson (Committee Member)
Lubomir Popov (Committee Member)
Generalizability of a risk perception measure is essential to make inferences and predictions based on the measured risk perception ratings. The present study used generalizability (G) theory to examine the generalizability of risk perception ratings across people. Archival data with risk perception ratings of samples from three different countries, United States, Japan, and China (total n = 617), were used for generalizability (G) and decision (D) studies based on G theory. For the combined sample, the G study results showed large variance across domain and small variance across people. But there was also considerable variance due to random errors, which may confound with the interaction effect between persons and activities in specific domains. Compared to the combined sample, the magnitudes of variance components were slightly different when examining the subsamples of males and females, and those from the United States, Japan, and China. D study results showed that sufficient generalizability could be achieved from risk perception ratings by a small number of raters, but this number vary across different population groups, and across different risk domains.
Wang, Yi, "Decomposing Variance Components for Risk Perceptions Using Generalizability Theory" (2017). Psychology Ph.D. Dissertations. 196.