Investigating the Role of the Additional Text on the Effectiveness of Impact Based Warnings

Last year, the National Weather Service (NWS), in its central region, implemented the first phase of a project designed to improve the quality and informativeness of severe weather warnings, with a special emphasis on tornado warnings. A main goal of these Impact Based Warnings (IBWs) is to improve the threat warning process and motivate appropriate responses by using event tags and additional textual material that provide more specificity about the magnitude of the storm and its potential consequences. Presentations at last year's AMS conference outlined some early initial success with the IBW program. Losego (2013) reported that forecaster's like the tiered system of tags that provide additional information about a tornado's intensity. Hudson et al. (2013) also reported some initial validation results comparing the use of the Tornado Threat tags to the actual occurrence of a tornado in terms of hits and misses. They reported a moderate degree of correspondence between IBW tornado warnings and an actual tornado event. Interestingly, however, others have criticized the IBW approach, with much of the dissatisfaction focused on cases where a tornado threat was listed as either considerable or catastrophic and for which an actual tornado did not occur (a miss). Some argue that these misses, in the face of a warning that appears more certain, simply undermines the credibly of all such warnings.

The research reported here represents an attempt to empirically investigate the effectiveness of the new IBW approach. The current study focuses on the actual content of the message, and whether or not the additional textual information in the IBW messages increases the perceived likelihood of an imminent event and the need to take protective action in the eyes of the public. Using a sample of undergraduates, who potentially would not be biased by knowing the distinction between IBW and non-IBW warnings, our participants read actual 2013 NWS warnings issued by the central region NWS offices participating in the IBW project. Some of the warnings were in their original version (i.e., they included the additional IBW information), while other warnings were missing the IBW information (equivalent to the old warning messages). Participants then rated their certainly on a 100 point scale that a tornado actually occurred, and we timed their responses as an indication of how easily they made their decision. We also asked for other information about message informativeness. If accepted, we will report the results of our study and discuss how they relate to the overall effectiveness of IBWs.