Table 3. Bivariate logistic regression of evacuation predictors on evacuation among all respondents and separately by level of education.

Voluntary shelter evacuation
Independent variables All respondents
(N = 212)
Illiterate
(N = 43)
Educated
1 to 10
(N = 127)
Educated
11 to 17
(N = 42)
Income and socioeconomic status
Income 1 1 1 1
Housing Quality 0.66 2.67 0.59 0.222***
Indicators related to access, understanding, and
interpretation of warning information
Severity of cyclone 2.42*** 2.43 3.13** 1.83
Landfall location of cyclone 2.61*** 2.55 2.91** 2.27
Landfall time of cyclone 2.35*** 2.55 2.66** 1.75
Evacuation order in the warning 1.51 1.41 1.83 1.5
Protective guidance in the warning 2.28*** 2.36 2.64** 1.75
Clarity of the warning message 2.36*** 3.74** 1.51 6.67**
Type of channel 2.036** 3.89* 1.74 1.93
Estimate of damage due to cyclone given a particular severity 1.73** 2.89* 1.58 1.25
The number in the cells are exponentiated betas or odds ratios. An odds ratio of 1 (or not significantly different from 1) indicates no effect. An odds ratio larger (or smaller) than 1 indicates factor by which odds of evacuation are increased (or decreased) by one unit change in independent variable. *, **, *** represent the level of significance (alpha)
*** implies alpha ≤ 0.01
** implies alpha ≤ 0.05
* implies alpha ≤ 0.10