Geographic scale and probabilistic forecasts: a trade-off for protective decisions?

Abstract

This pilot study aimed to examine the impact of varying geographic scales, probabilities of tornado occurrence, and presentation formats within severe weather forecasts on individuals' protective decisions. This pilot was unique in its specific examination of the tradeoff between highly localized geography and higher valued probabilistic threat information in weather-related decision-making. This pilot utilized a 4 (geographic scale) x 12 (probability) x 3 (forecast presentation format) mixed, nested experimental design. Participants were 440 United States adults who completed electronic questionnaires containing experimentally manipulated severe weather forecasts. A linear mixed model analysis revealed several findings. First, participants who saw only categorical forecasts made similar preparatory decisions across geographic scales. Additionally, they were more willing to take preparatory action as categorical risk increased. Second, when probabilities were presented, the propensity to take protective action was greater at higher probabilities and at larger geographic scales, affirming the regional geographic reference class selected by the Storm Prediction Center in today's outlook system. Third, individuals' propensity for action generally increased as scale and probability increased but the pattern varied across presentation formats. Lastly, participants reported having a map to look at was moderately important to their decisions and having probabilistic and categorical risk information was highly important to their decisions. Taken together, the findings suggest a complex relationship between geographic scale and probability, which is further complicated by forecast presentation format.

How Close is Close Enough? A Discussion of the Distances Relevant to Personalizing Tornado Risk

“…Results from a pair of studies suggests that scale may be more relevant when the event is real and personal to the participant. For example, a study focusing on real-world tornado warnings found that larger polygons led to people being less likely to take shelter (Nagele and Trainor 2012) while a study with hypothetical severe weather forecast scenarios found people were more willing to take preparatory action when the warning was impacting a multiregion area rather than a city (Shivers-Williams and Klockow-McClain 2021). The two studies are not directly comparable, but the differences suggest a need for future studies involving scale and local places with which participants are likely to be familiar.…”

“…We were not able to account for this in the study. Past research has shown that even telling participants that they live in a particular location may not be sufficient (Shivers-Williams and Klockow-McClain 2021). Research currently being conducted by the authors is examining whether county borders are relevant to descriptions of personalized risk areas.…”

“…Results regarding the influence of the geographic scale of information on tornado risk perception and protective action are mixed. The results of an experimental study using hypothetical severe weather forecasts similar to SPC convective outlooks suggested that people were more willing to take preparatory action when the geographic scale of the information was broader, such as impacting a multiregion area rather than a city, even if they were asked to imagine they lived in that city (Shivers-Williams and Klockow-McClain 2021). When using location information from actual tornado warning polygons, Nagele and Trainor (2012) found that larger polygons led to people being less likely to take shelter; when polygons covered more than 50% of the county, people were marginally less likely to seek additional information.…”

“You are here”: Assessing the inclusion of maps in a campus emergency alert system

“…Maps supplementing short mock alert messages have the potential to reduce confusion and increase understanding of emerging threats (Sutton & Fischer, 2020), and have been shown to improve comprehension and behavioural compliance intention in a crisis (Liu et al., 2017). Shivers‐Williams and Klockow‐McClain (2020) report maps are only moderately important in decision‐making but having a map to reference during decision‐making was evaluated as equally important by all participants regardless of the warning information they received.…”