Methods, Quantitative

Quantitative research methods in the study of crises are highly unstructured. Across multiple disciplines and methods of inquiry, there exists no agreed-upon set of procedures for the length of time of data collection, sampling techniques, and analyses of data. This may be attributed in part to the novel set of circumstances under which crises arise. Because crisis events tend to be unpredictable and exist outside a laboratory setting, adherence to strict scientific methods is often unrealistic. Therefore, in the study of crisis, the quantitative scientist must be sure to articulate his/her choices to bolster credibility of the methods used to collect, analyze, and interpret data.

Highly standardized or textbook methods for sampling and interacting with research participants, such as random sampling and the random assignment of individuals to stimulus and control groups, are often impossible during or in the immediate aftermath of a crisis. Take, for instance, the restrictions that are often associated with the physical location of a crisis. In some crises, the site of a disaster may be dangerous, and data collection may involve significant personal risk. Often, crisis research is conducted a significant length of time after the event, when conditions have normalized, requiring the respondents to answer questions about prior events, introducing the issues of bias, memory lapse, or any number of types of retrieval error or threats to validity. Moreover, eliciting participation from survivors of crises is often difficult. These factors, among others, promote the use of some type of survey methodology, usually self-administered, face to face, computer mediated, telephone, or otherwise assisted. In contrast, it is nearly impossible to produce true field experiment conditions in the context of a crisis or emergency, given their complex, unexpected, and often dangerous characteristics.

Researchers are presented with further challenges in terms of sampling. For a number of reasons, it may actually be advantageous for the crisis researcher to refrain from using randomized designs. An increasingly common preference in crisis research is to collect data quickly in order to minimize memory effects or recall heuristics. The more time between an event and administration of a questionnaire, the more likely the respondent is to use imprecise methods of recall, including mapping, heuristics, and decomposition. Thus, researchers are increasingly relying on [Page 619]nonrandomized data that are collected as quickly as possible after an event.