Public response to hurricane probability forecasts

"In the spring of 1983, an experiment was conducted to measure the effect which hurricane probability forecasts would have on public response. One group of respondents in Pinellas County, Florida was presented with 16 hypothetical hurricane threat situations described in terms of storm severity, storm location, National Hurricane Center "alert" (watch, warning, neither), and local officials' statements regarding evacuation (advised, ordered, neither). Another group of residents was presented with exactly the same 16 threat situations, plus the probability that the storm would affect their area and the probabilities of its affecting other coastal locations. People in both groups were asked whether they would evacuate in each of the 16 situations. The 16 threats were constructed such that the variables involved (severity, NHC alert, etc.) were statistically independent of one another. Overall, probabilities were found to have little, if any, effect on public response one way or the other. People clearly compared their probability of being affected to the probability of other locations being affected. If people perceived their probability to be notably higher than others, evacuation rate was enhanced slightly compared to the no probability situation. if they perceived their probability to be notably lower than others, evacuation rate was reduced slightly compared to the no probability situation. By far the most important variable affecting response was local officials' statements, regardless of whether probability information was available or not."