Cultural theory and cultural cognition theory survey measures: confirmatory factoring and predictive validity of factor scores for judged risk

Abstract

Two survey approaches to measuring cultural effects on risk views have been developed, both informed by the grid/group approach of Douglas and Wildavsky, well known to many risk analysts. Using data from an online national US panel, confirmatory factor analyses were conducted of ‘cultural theory’ (CT) and ‘cultural cognition theory’ (CCT) measures, which respectively purport to measure (1) four worldviews or cultural biases – hierarchism, individualism, egalitarianism, and fatalism – using both individual items and longer statements, and (2) indices of grid (hierarchism-egalitarianism) and group (individualism-communitarianism) using individual items. Confirmatory factor analyses showed CT measures (items and statements) tended to cluster as expected (i.e. four worldviews), particularly items which on their own exhibited the best fit, while CCT items loaded on the two hypothesized dimensions but with poor fit (particularly the reversed egalitarian and communitarian items to represent high-grid and low-group, respectively). Combining all measures representing the four cultural biases, or adding in the communitarian subscale for CCT, also created plausibly coherent factoring (e.g. hierarchical measures from CT and CCT loaded together, as did the CT-fatalist items and statements with the CCT-communitarian items), but model fit was poor again. Depending upon the criterion used, CT or CCT items’ factor scores appeared the superior predictor of personal and U.S. risk perceptions, but overall predictive validity was modest. These results were somewhat improved among those with high (versus low) political knowledge. If survey-based research on culture’s role in risk views is to thrive, further scholarly attention to such measurement issues is warranted.