Factors Affecting General Public's Protective Behavior during May 2013 Tornado Outbreaks in Oklahoma

Factors Affecting General Public’s Protective Behavior during May 2013 Tornado Outbreaks in Oklahoma

Seyed M. Miran¹ Chen Ling¹ Lans Rothfusz²

¹Department of Mechanical Engineering, University of Akron, Akron, OH

²NOAA National Severe Storms Laboratory, Norman, OK

Tornado occurs in the United States more than any other countries in the world and causes casualty and property damages every year.  To minimize its impact, it is crucial that people take protective actions. On May 19^th, 20^th and 31^st of 2013, three tornados with different intensities struck Oklahoma City metropolitan area. Two month later, survey data were collected from 124 residents regarding their protective behaviors after each of the tornado events. Based on this data set, this study aims to explore the influential factors on people’s protective actions, including environmental context factor (proximity to the tornado), information factor (number of weather information sources ) as well as personal factors (age, marital status, gender, education, experience from the past tornados, type of geographical area- rural vs. urban-, income).   People’s protective actions were regressed on the above mentioned factors using logit models for each of the three tornado events to figure out which of those factors were statistically significant.

The result of the logistic regression model showed that factors of “number of weather information sources” and “proximity to the tornado” in models of all three events were statistically significant and had positive association with the likelihood of taking protective actions.  Additionally, for the May 19^th model, the “experience” and “education” were significant factors but for those of May 20^th and May 31^st, neither factors played a significant role in taking protective actions. It is possible that people were more vigilant after the previous tornado.

Tornado watch, tornado warning and tornado siren are three types of authoritative alerts that were sent out on all three tornado events. Besides the regression analysis, in order to investigate the effect of alert type on the protective actions across three events, Pearson’s Chi-Square Test was used to analyze participants’ behaviors after receiving each of the alerts. No significant results were found which showed that people reacted consistently in response to the alerts across three tornado events. Siren posed more fear to people and caused more people to take shelter than tornado warning, and tornado warning poses more fear than the tornado watch.