Natural Hazards and Job Choice: Factors driving job choice and willingness to relocate

We examined all factors related to the job characteristics to understand the drivers of the final job choices by participants. The factors included the salary, the crime risk and hazard risk of the job’s location, the change in hazard risk between where the participant was currently living and the job’s location, the geographic distance (in terms of US Census divisions) the participant would move, and the change in base cost of living between the participant’s current location and job location.

Salary was the strongest driver of job choice, showing the highest statistical significance (p < 0.001) and a positive association, indicating that higher-paying jobs were more likely to be chosen. Crime risk was also a significant factor (p < 0.001), with a negative association, suggesting that jobs located in areas with higher crime risk were less likely to be selected. Hazard risk showed a marginally significant negative effect (p = 0.051), implying that higher hazard risk may slightly reduce the likelihood of job selection, though this effect was weaker than salary or crime risk. Relocation factors, such as the distance between the job and the respondent’s current location (measured by US Census division change), and the change in hazard risk, did not significantly influence final job choice. Similarly, the cost of living, both as an absolute measure and as a change relative to current location, had no meaningful impact on the final decision.

When comparing geoscientists and non-geoscientists separately, salary remained the strongest and most consistent predictor across both cohorts. Crime risk also had a consistently negative and significant effect in both cohorts. Hazard risk was marginally significant only for geoscientists, suggesting that they may be more sensitive to choosing a job located in higher hazard risk areas. No other factors significantly influenced job selection in either group.

Whole Population — Logit Regression Summary

Geoscientists Only — Logit Regression Summary

Non-Geoscientists Only — Logit Regression Summary

Is Job Choice Behavior Different Between Cohorts?

To determine whether final job choice behavior differed between geoscientists and non-geoscientists, we conducted a series of statistical comparisons using three approaches: coefficient-wise z-tests, interaction term testing, and a likelihood ratio test (LRT).

First, we ran separate logistic regression models for each subgroup and compared the resulting coefficients using z-tests. None of the differences in coefficients across key predictors such as salary, hazard risk, crime risk, and cost of living were statistically significant (all p > 0.30).

Coefficient Comparison — Geoscientists vs. Non-Geoscientists

Next, we estimated a full interaction model including geoscientist status (geo) and interaction terms between group membership and each predictor. A Wald test on each interaction term revealed no significant group moderation effects (all p > 0.30), indicating that the influence of predictors on job choice did not significantly differ by group.

Interaction Term Tests — Geoscientist × Variable

Finally, we conducted a likelihood ratio test comparing the full interaction model to a simpler pooled model with no interaction terms. The result (LR statistic = 3.76, df = 7, p = 0.81) confirmed that the interaction model did not provide a significantly better fit to the data.

Taken together, these results demonstrate that geoscientists and non-geoscientists respond similarly to the job characteristics examined. Therefore, for all subsequent analyses, we use a single pooled model including the full sample. This approach preserves statistical power and simplifies interpretation, without compromising accuracy.

Does Disposable Income Factor into Job Choice?

We calculated disposable income for participant’s final job choice as final job salary minus the cost of living for the location of the job to understand how strongly purchasing power factored into job choice. Disposable income was a strong positive predictor of job selection (p < 0.001), indicating that jobs offering higher earnings relative to cost of living were more likely to be chosen.

Whole Population — Logit Regression

Does Willingness to Relocate Influence Job Choice?

We examined whether a participant’s willingness to relocate influenced job choice behavior. The willingness to move variable represented individuals preferred geographic job search scope (i.e., only within their current US Census division, both within and outside of their current US Census division, only outside of their current US Census division). The coefficient for willingness to move was negative but not statistically significant, indicating that individuals’ openness to relocating did not significantly affect the likelihood of choosing a particular job once job characteristics were accounted for.

Whole Population — Logit Regression

We also examined whether an individual’s willingness to move affected the likelihood that they would choose a job that’s in a different US Census division than their current location of residence (division change). Neither willingness to move nor the interaction with division change were statistically significant, suggesting that individuals’ openness to relocating did not significantly moderate their likelihood of selecting jobs located outside their current region, after accounting for job attributes.

Whole Population — Logit Regression Summary

To evaluate whether the effect of salary on job selection varies by willingness to relocate, we included an interaction term between willingness to move and salary in the logistic regression model. The interaction between salary and willingness to move was not statistically, nor was the main effect of willingness to move. These findings suggest that individuals’ stated openness to relocation does not meaningfully alter how salary influences their final job choice. There was no evidence that more mobile individuals were more responsive to higher-paying jobs. As such, stated willingness to move had no independent or moderating effect on job selection once job-level features were accounted for.  

Whole Population — Logit Regression

Continue reading

• About the game
• Methodology
• About the participants
• Hazard exposure and concern
• Factors influencing job choice and relocation
• Reasons for current location, occupation, and final job choice
• Alignment of attitudes and actions
• Predicting job choice outcomes
• Conclusions