Date of Award

January 2023

Document Type


Degree Name

Doctor of Philosophy (PhD)


Aerospace Sciences

First Advisor

Mark J. Dusenbury


The United States Coast Guard (USCG) implemented an aviation safety survey prior to this research (ver. 0). Cooley (2019) uncovered opportunities to improve the survey’s validity, creating a psychometrically-sound safety climate instrument. Inconsistencies with what and how to measure safety climates still exist in the corpus of literature. More attention is needed on safety management systems’ (SMS) predictive metrics, particularly for Coast Guard aviation outfits throughout the world. This research study used an exploratory sequential mixed methods design followed by an additional phase of quantitative research methods. Extensive deliberation with USCG stakeholders produced survey research questions, guiding the survey design (ver. 1). After one survey campaign, an exploratory factor analysis (EFA) was conducted, resulting in a second version of the safety climate assessment instrument (ver. 2). T-Tests and ANOVAs were performed to determine differences in survey demographics. A confirmatory factor analysis (CFA) was conducted after the second survey campaign to verify the survey validity. A MANOVA test was conducted to examine the effect of survey ver. 2’s demographics (independent variables) on the survey’s constructs (the dependent variables). Survey constructs were then used as independent and dependent variables for regression analyses. Lastly, multiple regressions were conducted with survey constructs and mishap data to address hypotheses that positive survey responses are commensurate with low mishap rates. The EFA results suggested a consolidation from eight to five aviation constructs, as well as consolidating items that were split based on demographic. T-Tests of ver. 1 demographics indicated significant differences between the highest and lowest scoring groups per analyzed demographic. An ANOVA was conducted on the demographic with the most groups, Air Stations (units). Approximately 2/3 of the air stations differed significantly from the others. CFA results suggest that the five-construct model of ver. 2 had better fit indices than consolidating all survey items onto a single construct. Further, ver. 2 could be improved upon by removing survey items, itemized in Chapter 3. From the MANOVA, the survey’s demographics had a significant main effect on the survey’s constructs. For regression analyses, the Just & Reporting Culture (JRC) and Safety Leadership constructs had significant positive effects on the Risk Management (RM) construct. The interaction effect between JRC and Safety Leadership was also significant. Statistical significance varied between the survey and mishap data; RM had significance, with a negative relationship indicating that as RM is scored higher on the survey (per unit and per asset), mishap counts decreased. RM and JRC significantly predicted total mishap counts, human factors-related mishap counts, and Operational Hazard Reports; RM had a positive relationship, while JRC had a negative relationship.