Assessing the Influence of Impact Based Tornado Warnings on Shelter in Place Decisions

The National Weather Service recently implemented nation-wide enhanced tornado warnings known as Impact Based Warnings (IBWs). The overarching goal of these IBWs is to improve the threat warning process and motivate appropriate responses by using more specific text about a tornado’s potential impact and severity, including potential consequences. Research recently published by the author in Weather, Climate, and Society (and reported at the AMS meetings in both 2014 and 2015) found that the IBWs did indeed promote more shelter in place decisions compared to non-IBWs. In that research, undergraduates adopted the role of a plant manager and read both IBW and non-IBW warnings line-by-line. At three different decision locations, participants made decisions about having the plant shut down and employees shelter in place. My results showed that the IBW warnings produced higher likelihoods of closing the plant and sheltering in place, but only after the additional IBW information was presented, supporting the effectiveness of the IBW approach.

The current research presented here expands upon my previous work in three main ways. First, participants made shelter in place decisions to tornadoes of potentially much greater magnitude (EF3-EF4). These IBWs also included the following threat tag: “TORNADO DAMAGE THREAT. . . CONSIDERABLE.” This manipulation was included to see if the presence of the threat tags (which were not included in the previous research) increased shelter in place decisions. Second, participants were allowed to read each tornado warning in its entirety, rather than line-by-line as in my earlier research. One methodological question of interest is whether my previous use of line-by-line reading forced participants to read the weather warnings in a more careful way than they would have otherwise. Third, the participants were much more diverse with respect to a variety of demographic variables having been recruited through a nationwide pool. These last two changes were specifically implemented with the goal of increasing the research’s ecological validity. With these changes in mind, the rest of the experiment’s methodology was identical to that used previously – participants read the full tornado warnings and at three different decision locations, using a 0-100 likelihood scale, made decisions about having the manufacturing plant shut down and employees shelter in place. Results will be compared to those obtained previously and discussed within the context of NWS best practices.