Positive relationship between education level and risk perception and behavioral response: A machine learning approach

PLOS ONE | https://doi.org/10.1371/journal.pone.0321153 April 3, 2025 8 / 10 PLOS ONE Positive relationship between education level and risk perception and behavioral response feature. Each point represents three elements: the vertical axis indicates the feature, the color reflects the feature value (with red representing higher values and blue lower values), and the horizontal axis shows the direction of influence on the predicted value. Based on the feature importance ranking and the impact of each feature on preparatory behavior, the following conclusion can be drawn. The education level plays a crucial role in predicting behavioral intention and serves as a significant contextual factor. Although its explanatory power is somewhat reduced in the behavioral intention model, it remains moderate. Unlike many variables, which cluster around a SHAP value of zero, the distribution of education level shows distinct patterns. 3.3 Examination of contextual evidence Based on ML model results, education level (Q12) emerges as a significant contextual factor. To validate this, a statistical test is conducted to assess its impact on preparatory behavior. The effect size is analyzed in two groups: one with education level as a control variable and one without. In the group where education level is included, the average effect size is 0.0733; in the group without it, the average effect size is 0.0764. Given education’s potential importance in specific contexts, we hypothesize a significant difference between these groups, which is tested using a t-test.

Before performing the t-test, it is essential to test for variance homogeneity, as this is a prerequisite. The homogeneity test shows a significant difference in variances between the two groups, statistically significant at the 1% level. Therefore, we use the t-test for unequal variances. The results indicate that the mean effect size in the group without education level as a control is significantly higher than in the group with it, with statistical significance at the 1% level. This strongly supports the conclusion that education level plays a crucial role in the relationship between risk perception and preparatory behavior. To assess the impact of risk perception on behavioral intention, the original evidence is divided into two groups: one that includes education level as a control variable and one that does not. The group considering education level has a mean effect size of 0.1727, while the other group has a mean effect size of 0.1725. If education level is a significant contextual fac- tor, a notable difference should exist between the two groups. A t-test is conducted to examine this. Prior to the t-test, the homogeneity of variances must be verified. The results indicate a significant variances difference at the 1% level, prompting the use of a t-test for unequal vari- ances. The findings show that the mean effect size with educating level as a control is signifi- cantly higher than without it. 4. Discussion In summary, education level is a critical contextual factor for several reasons. First, individu-

als with higher education typically possess enhanced knowledge and information-processing skills [25], enabling them to better understand and assess risks, leading to more effective preparatory measures. Second, education strengthens risk awareness, with more educated individuals being more vigilant and proactive in preparing for potential risks [26]. Third, higher education fosters logical thinking and critical analysis, allowing individu- als to assess situations calmly, develop targeted response strategies, and implement effective preparatory behaviors. Finally, individuals with higher education are more adept at acquiring and utilizing resources, leveraging information, expert advice, and social networks to manage risks.