Individual perceptions of flood and COVID-19 risks at the start of the 2020 hurricane season
A comparison of perceptions of COVID-19 and hurricane-related risks shows that perceptions of COVID-19 risks at the start of the 2020 hurricane season exceed those of flood risks. For example, more people are worried or strongly worried about COVID-19 risks (63%) than about flood risks (33%), as Fig. 2 illustrates. Moreover, only 21% disagree and 7% strongly disagree with the statement, “The probability of flooding is so low that I am not concerned about the consequences of a flood.” These percentages are 28% and 30%, respectively, for a similar statement about the consequences of being infected by the coronavirus. Individuals not only worry about the negative consequences of COVID-19, but also perceive high infection risks. For instance, 34% of respondents believe that it is likely or very likely they will become infected by COVID-19, and 39% expect to become very ill or extremely ill, once infected. More than half of the sample already experienced expenses because of COVID-19, mainly due to a loss of income. The vast majority of respondents (81%) are worried about the current economic situation.Footnote 6 Previous research has shown that feelings toward risks are likely to affect how people prepare for a disaster. Examples are the worry about the consequences of a hazard or perceptions about whether or not the probability of experiencing a threat is high enough to trigger concern (Kunreuther and Pauly 2018; Botzen et al. 2019). In summary, our survey results indicate that the start of the 2020 hurricane season is dominated by concerns about COVID-19, which can influence hurricane preparedness activities.
Evacuation intentions at the start of the 2020 hurricane season
When being asked about intentions to evacuate to a safer place under a voluntary evacuation order at the start of the hurricane season, 39% of respondents answer that it is likely or extremely likely they would evacuate (see Table 2).
Descriptive statistics indicate that concern about COVID-19 is the most important obstacle for evacuation during the 2020 hurricane season. Both of our surveys in February and June 2020 contained a question about the obstacles for evacuation during a hurricane threat. More respondents indicated at least one potential obstacle in the June survey than in the February survey (75% versus 56%). The inability to pay for hotel costs was the most frequently mentioned obstacle during Hurricane Dorian (by 26% of the February 2020 survey respondents who had an obstacle). However, as Fig. 3 illustrates, hotel costs dropped to the number four obstacle during the 2020 hurricane season, although the percentage of respondents who list hotel costs as an obstacle remains stable at 26%. Instead, COVID-19 was mentioned the most frequently, by almost half of the respondents to the June survey who expected to experience any obstacles. Although these results were obtained from two different data collection methods and samples, they are indicative that COVID-19 became an important obstacle for evacuation during the pandemic.
We conduct a series of statistical analyses to examine how evacuation intentions under a voluntary order depend on socio-demographic characteristics and perceptions of the hurricane and COVID-19 risks. An ordered probit model of the intentions to evacuate voluntarily with only socio-demographic characteristics as explanatory variables finds that older people are significantly less likely to evacuate (Table 3).Footnote 7 As a next step, we add perceptions of the hurricane and COVID-19 risks and length of residence as explanatory variables to the model to examine whether the significant relationship between age and intentions to evacuate still holds once these explanatory variables are controlled for. We include the variables: length of residence, perceived flood probability, worry about flooding, perceived coronavirus infection probability and concern about COVID-19 in model 2, since these variables are all significantly correlated with age (Pearson correlation coefficient p values < 0.05), and are potential predictors of voluntary evacuation intentions. Therefore, these variables may explain some of the relationship between age and voluntary evacuation intentions. Variables that are uncorrelated with age are omitted from the regression in model 2. In a mediation analysis (Table 4), we include only significant mediating variables when calculating indirect effects, which is the share of the relationship between age and voluntary evacuation that can be attributed to mediating variables. The other potential mediators are controlled for in a model that includes control variables.
We find that the likelihood of voluntary evacuation significantly increases with worry about flooding, but significantly declines with concern about the consequences of becoming infected by COVID-19 and the length of residence (Table 3).Footnote 8 Moreover, the independent effect of age on intentions to evacuate voluntarily becomes insignificant, indicating that the significant negative effect of age in the first model is an indirect effect, perhaps driven by perceptions of flood and COVID-19 risks as well as the length of residence. This result is examined in more detail using a mediation model (Table 4).Footnote 9
Table 4 displays the total effect of age on voluntary evacuation, divided into a direct and indirect effect via concern about the consequences of becoming infected by COVID-19, worry about flooding, and length of residence. Overall, the total effect shows that older individuals have lower evacuation intentions. Controlling for concern about the consequences of becoming infected by COVID-19, worry about flooding, and length of residence leaves an insignificant direct effect of age. The indirect effect, which is the share of the relationship between age and voluntary evacuation that can be attributed to perceptions of COVID-19 and flood risks and length of residence, is explained by the coefficient estimate − 0.007 (p value < 0.01). Between 60 and 73% (depending on included control variables) of the relationship between age and voluntary evacuation is explained by concern about the consequences of becoming infected by COVID-19, worry about flooding, and length of residence.Footnote 10 The two risk perception variables are statistically significant and explain a larger proportion of the relationship than the length of residence.
Evacuation intentions during Hurricane Eta
When respondents to our real-time survey during the threat of Hurricane Eta were asked when they were going to evacuate to a safer place, 35% answered this is very unlikely, 27% answered unlikely, 10% answered likely, and only 6% answered very likely. We repeated the same analyses of evacuation intentions at the start of the 2020 hurricane season (that are reported in Tables 3 and 4) for evacuation intentions during Hurricane Eta, which hit Florida at the end of the hurricane season in November 2020. These results for evacuation during Hurricane Eta are reported in Tables 5 and 6. The ordered probit model results in Table 5 confirm our previous findings that evacuation intentions are negatively related to age (model 1), of which the significance declines to marginally significant in model 2 when risk perceptions are added. These findings again show that evacuation intentions are negatively related to concern about the consequences of becoming infected by COVID-19, and positively related to flood risk perceptions. Furthermore, although the sign of the coefficient estimate on the length of residence is the same in Table 5 as Table 3, this estimate is not significant in Table 5. Whereas, the perceived coronavirus infection probability is significantly positively related to evacuation intentions, which may be due to people with higher intentions to evacuate perceiving that they are more likely to become infected by COVID-19 in the event that an evacuation is in fact ordered.Footnote 11
Moreover, the mediation analysis results in Table 6 confirm that the main pattern of findings that are reported in Table 4 for evacuation intentions at the start of the 2020 hurricane season also hold for evacuation intentions among a separate sample who faced Hurricane Eta at the end of the 2020 hurricane season. More specifically, a large proportion of the relationship between age and voluntary evacuation intentions (in this case 42%) is explained by concern about the consequences of becoming infected by COVID-19 and worry about flooding. However, worry about flooding is only a statistically significant mediator without other control variables added to the model. Length of residence is also an insignificant mediator in Table 6 which is expected given the lack of significance of this variable in Tables 4 and 5.