Date of Graduation
9-2025
Document Type
Thesis
Degree Name
Master of Science in Statistics and Analytics (MS)
Degree Level
Graduate
Department
Statistics and Analytics
Advisor/Mentor
Robinson, Samantha
Committee Member
Giovanni Petris
Second Committee Member
Qingyang Zhang
Keywords
Conditional Inference Trees; Decision Trees; Fatalism; Geographically Weighted Regression; Psychological Constructs; Spatial
Abstract
Psychological constructs such as Anxiety, Depression, Fatalism, Divine Control, Luck, Helplessness, and Internality are pressing subjects in the United States (US). Although several studies have explored how specific covariates influence these constructs, less is known about how combinations of predictors interact to predict these constructs and how these relationships vary across different geographic regions. This study addresses the gap by combining Conditional Inference Trees (CIT) and Geographically Weighted Regression (GWR) to explore the interaction effects of the covariates in predicting the constructs and spatial variability in the determinants of psychological constructs respectively Using a nationally representative survey of 2000 respondents, CIT was used to explore how covariates such as Adverse Childhood Experience (ACE), age, gender, race, education, and urbanicity interact to predict each construct. Geographically Weighted Regression (GWR) was employed to explore the spatial variation in the effects of covariates on psychological constructs in the U.S. Key findings reveal that age and Adverse Childhood Experiences (ACE) were strong predictors of different levels of Depression and Anxiety. Education, age, and race played critical roles in predicting different levels of Fatalism. Regional patterns also emerged: for example, education and urbanicity were more influential on Anxiety in the West and Central U.S., while gender had stronger effects in the Northeast and Midwest. Urbanicity more strongly affected Depression in the Northeast and Midwest, whereas gender was more impactful in the West and Central regions. Similarly, education and urbanicity were linked to Luck in the Northeast and Midwest, while ACE was more predictive in the West and Central U.S. Helplessness was associated with gender in the Northeast and Midwest and with race in the West. Education influenced Internality in the West and Central U.S., while gender was more impactful in the North and Midwest. This study highlights the importance of considering both interactions among mixed-type covariates and geographic variability when examining psychological constructs. By incorporating CIT and GWR, we were able to discover significant interaction between covariates in predicting psychological constructs and regional differences in how covariates such as education, gender, race, and urbanicity influence constructs like Anxiety, Depression, Fatalism, and Divine Control.
Citation
Agyei-Owusu, F. A. (2025). Differential Drivers of Psychological Constructs: A Comparative Analysis Using Conditional Inference Tree and Spatially Varying Coefficient Models. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/5848
