Disastrous Inferences? The Ecological Fallacy in Disaster and Emergency Management Research
Forthcoming in Jason D. Rivera, ed., Research Methods of Disaster and Emergency Management, Routledge 2021, Forthcoming
35 Pages Posted: 25 Sep 2020
Date Written: July 11, 2020
In research on disaster and emergency management, the limited available data is often summarized at higher levels of aggregation, particularly summaries for geographic areas – sometimes referred to as “ecological data.” Measures of social vulnerability and disaster post-mortems based on such data purporting to measure or explain individual-level phenomena have been criticized as committing an “ecological fallacy” (Beccari 2016; Duneier 2006). Stated briefly, one commits an ecological fallacy if one assumes that relationships observed at an aggregate level imply that the same relationships exist at the individual level. In fact, estimates of causal effects from aggregate data can be wrong both in magnitude and direction (Robinson 1950). This article examines the factors that contribute to incorrect inferences that can arise from the analysis of aggregate data and provides guidance to disaster and emergency management researchers to help them avoid committing ecological fallacies.
Keywords: ecological fallacy, bias, research methods, aggregation
JEL Classification: C18, C21, C23
Suggested Citation: Suggested Citation