When Does Analytic Complexity Matter? A Theory-Informed Comparison of Regression-Based and Algorithmically Flexible Models of Student Success in Community Colleges

Student persistence and academic performance are central indicators of student success in higher education, yet they are often modeled using similar analytic assumptions despite reflecting distinct underlying processes. Guided by theories of student engagement and persistence, this study examines whether student persistence intention and college grade point average (GPA) differ in their structural complexity and whether analytic approaches of varying complexity offer differential explanatory value. Using data from 23,215 students in the 2024 Community College Survey of Student Engagement (CCSSE) in the United States, we compare pre-specified additive regression models, regularized regression models, and algorithmically flexible tree-based models to predict persistence intention (self-reported likelihood of reenrollment) and GPA using the same theory-aligned predictors. Results show that regression-based approaches performed as well as algorithmically flexible models for persistence intention and better than them for GPA, and that no model exceeded the majority-class baseline in classification accuracy for persistence intention. Theory-informed interaction terms suggested by the tree-based models did not improve fit or prediction when specified directly within regression, and a multiple-imputation sensitivity analysis indicated that substantive estimates were robust to the missing data produced by listwise deletion. These findings indicate that analytic complexity offered no predictive gains for either outcome, and that apparent context-dependent relationships require explicit specification and testing rather than being inferred from algorithmic flexibility alone. By comparing analytic approaches across two theoretically distinct outcomes, this study highlights the importance of aligning methodological choices with outcome structure and contributes to theory-informed research on student success in community colleges.