Cross-sectional studies of attitude-behavior relationships are vulnerable to the inflation of correlations by common method variance (CMV). Here, a model is presented that allows partial correlation analysis to adjust the observed correlations for CMV contamination and determine if conclusions about the statistical and practical significance of a predictor have been influenced by the presence of CMV. This method also suggests procedures for designing questionnaires to increase the precision of this adjustment.

This study collected data on the evacuation from Hurricane Lili to answer questions about households’ reliance on information sources, the factors affecting their decisions to evacuate, the timing of their hurricane evacuation decisions, and the time it took them to prepare to evacuate. The results replicated previous findings on the sources of hazard information, evacuation concerns, and the timing of evacuation decisions. In addition, they provide new information about evacuation preparation times and the finding that household characteristics are uncorrelated with evacuation decision times or evacuation preparation times.

Research on the community impacts of natural disasters has yielded a wide variety of findings, but no coherent model of the process by which hazard agent characteristics produce physical and social impacts. This article summarizes the principal features of this process and describes the ways in which hazard mitigation and emergency preparedness practices can limit the physical impacts and the ways in which community recovery resources and extra-community assistance can reduce social impacts.