Tornado Safety at Night: Understanding Public Receipt of and Response to Warnings

Background and Purpose. Tornadoes at night are twice as likely to have a fatality as those during the day, with physical and emotional trauma for affected families and communities. While tornado warning receipt is an important factor in models of whether an individual takes protective action (e.g., seeks shelter), nighttime warning receipt has not been studied. Also, understanding of the factors that lead some people, but not others, to take action after a warning is still limited and with mixed findings for variables such as gender, age, income, and race. The role of cognitive factors and the complexity of decision making that people face are often overlooked. To help reduce nighttime tornado fatalities, we examine who receives and how people respond to warnings in Tennessee—a state with a high risk of nighttime tornado fatality and in a region (Southeastern U.S.) with high frequency of strong tornadoes.

Methods. Data are from randomly-sampled phone surveys (N=1,804) of residents in 12 Tennessee counties. Participants were randomly assigned to day or night versions of the same instrument. We selected counties for their geographic and socioeconomic diversity. Survey measures included chance of receiving a warning once issued; intended actions after warning receipt (e.g., turn on the television, search the web, contact others, check outside, etc.); and several demographic and cognitive factors as potential predictors. Bivariate analyses and logistic regression compared chance and predictors of warning receipt for day versus night scenarios. Latent class analyses looked for subgroups of responders (i.e., people with shared patterns of what they do upon receiving a warning).

Results. About 84% of participants asked about a daytime tornado said there was a high/very high chance of receiving the warning, compared to 48% of participants asked about nighttime (U=241,285.50, p=.000). While demographic and cognitive factors helped predict daytime receipt, cognitive factors were salient at night. Perceived county risk (OR=1.174, p=.001) and prior tornado experience (OR=1.587, p=.003) were positively associated with chance of nighttime receipt, while belief that luck is important in survival (OR=0.844, p=.039) was negatively associated. Latent class analyses identified three subgroups for day and night: people who “carry on” with what they were doing (17% day, 4% night), “television traditionalists” who rely on local television for decision making (54% day, 68% night), and the “technology savvy” who use social media in decisions (29% day, 28% night).

Conclusions and Implications. Findings suggest that people are much less likely to receive an issued nighttime warning, that nighttime receipt may be more affected by cognitive than demographic factors, and that people can be classified into subgroups of responders. Further research on the identified subgroups can help the National Weather Service (NWS) and their media and other partners strategize on how to influence groups who are not responding to warnings. In collaboration with the NWS and partners, future research should test strategies that increase warning access at night and improve people’s response to warnings in ways that consider the multiple and diverse behavioral steps they may take before deciding to seek shelter.