HOUSEHOLD ADJUSTMENT TO EARTHQUAKE HAZARD A Review of Research

Data from 23 studies confirm theoretical predictions that households’ adoption of earthquake hazard adjustments is correlated with their perceptions of the hazard and alternative adjustments, demographic characteristics, and social influences. However, some findings require modification of existing theories of hazard adjustment. Examination of the methods used in previous investigations underscores a need for better theories, more complete testing of existing theories, and improved data analytic and data reporting procedures in future tests of those theories. Seismic risk has become an increasing concern since the 1971 San Fernando earthquake. Government at all levels has attempted to reduce vulnerability, but households also must act to limit casualties, property damage, and social/economic disruption. One obstacle to improving seismic safety has been a limited understanding of the process by which households decrease vulnerability. This process, hazard adjustment, encompasses actions that intentionally or unintentionally reduce risk from extreme events in the 461 AUTHORS’ NOTE: This work was supported by the National Science Foundation under Grant BCS-9633594. None of the conclusions expressed here necessarily reflect views other than those of the authors. Correspondence should be directed to Michael K. Lindell, Hazard Reduction & Recovery Center, Texas A&M University, College Station, TX 77843-3137. ENVIRONMENT AND BEHAVIOR, Vol. 32 No. 4, July 2000 461-501 © 2000 Sage Publications, Inc. © 2000 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. at PENNSYLVANIA STATE UNIV on April 14, 2008 http://eab.sagepub.com Downloaded from natural environment (Burton, Kates, & White, 1978; Moore, 1964). Preimpact adjustments include hazard mitigation, emergency preparedness, and insurance purchase. Hazard mitigation provides passive protection at impact (e.g., strapping water heaters to walls before an earthquake prevents property destruction). Emergency preparedness supports active response after impact (e.g., establishing supplies of bottled water and canned food before an earthquake allows people to survive disruption to food distribution). Insurance purchase redistributes the financial impact of damage across time and persons. Many studies in the past 25 years have related respondents’ adoption of seismic hazard adjustments to risk perception, demographic characteristics, personal experience, social influence, and other variables. The theoretical constructs, measured variables, and research designs of those studies have varied considerably. The profusion of approaches has yielded a wealth of new and useful ideas, but the idiosyncratic nature of many studies has impeded summarization of this work. A search of psychological and sociological abstracts identified 23 English-language studies published between 1974 and 1998 that correlated household seismic adjustments with other variables or reported data permitting such correlations to be determined (see Table 1). The next section examines these studies in terms of four methodological issues—vulnerability to random and systematic sampling errors and to random and systematic response errors. This section is followed by a summary of empirical findings regarding four classes of variables—risk perceptions, perceived adjustment attributes, demographic characteristics, and other variables. The empirical findings are followed by an evaluation of the congruence between these empirical findings and theoretical models and suggestions for future research. METHODOLOGICAL ISSUES Random sampling errors. The magnitude of random sampling errors in these studies generally is small because, as Table 1 indicates, only six studies had as few as 100 to 250 respondents, three had sample sizes of 251 to 500, and the remaining 14 studies each had more than 500 respondents. Statistical power analysis (e.g., Bailey, 1971) shows that even the smallest of these studies has excellent power (π = .95) to detect a population value of r = .20, whereas studies with N > 400 have excellent power to detect a population 462 ENVIRONMENT AND BEHAVIOR / July 2000 (text continues on p. 468) © 2000 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. at PENNSYLVANIA STATE UNIV on April 14, 2008 http://eab.sagepub.com Downloaded from 63 TABLE 1 Summary of Study Authors, Location, Sample Size, Design, and Reported Correlates of Seismic Hazard Adjustment Study Location Study Demographic Risk Adjustment Other Seismic Study (sample size) Design Characteristics Perceptions Attributes Variables Adjustments Jackson and San Francisco CrossExpected Spontaneously Spontaneously Mukerjee (1974) (n = 120 residents) sectional damage mentioned mentioned survey attributes adjustments Jackson Los Angeles, CA; CrossEducation, Expected Prior losses, Spontaneously (1977, 1981) Anchorage, AK; sectional income damage (ns) hazard salience mentioned Vancouver and survey adjustments