Using Tree-Based Models to Identify Factors Contributing to Trait Negative Affect in Adults With and Without Major Depression

Abstract

Abstract Background Individuals with high levels of negative affect (NA) are at an increased risk of experiencing distress and negative self-views. Theoretical models suggest that NA plays a critical role in psychopathology, particularly in Major Depressive Disorder (MDD), and is linked to cognitive-perceptual and affective regulation issues. Objective Determine whether maladaptive cognitive schemas, attributional style, childhood adversity, and lifestyle factors (including alcohol and drug use and physical activity) could effectively predict negative affect (NA) in adults. Methods A secondary data analysis was performed on a sample of 342 depressed and non-depressed adults. Beta regression and regression tree analyses were conducted to identify the principal risk factors and their interactions. The regression tree model was trained with 5-fold cross-validation on 75% of the sample, with 25% of observations held for testing. Results The findings revealed that the cognitive schemas of disconnection and rejection and impaired autonomy had a significant impact on the likelihood of higher scores on the State Depression Inventory (IDER) test (p < 0.001), as indicated by both beta regression and regression tree analyses. Additionally, childhood adversity emerged as a crucial factor in determining high levels of NA. The regression tree model achieved strong performance metrics, including an R-squared value of 0.77. Conclusions This study represents a significant step forward in the understanding of NA, as it considers a broad range of individual factors, such as cognitive schemas, lifestyle, and demographics, to predict its impact on NA, with potential implications for prevention programs aimed at reducing NA.