ABSTRACT
A wealth of research has linked perceptions of stress to both psychological well-being and physical health. What is less well understood are the circumstances that contribute to perceptions that life is overwhelming, uncontrollable, and considered stressful. Although theoretical accounts describe complex roles, experiences, and demands inherent to adult life, these life events and chronic stressors are rarely examined in concert. Furthermore, the diversity that exists among middle-aged and older adults suggests that concentrating on a single predictor or pathway may be short sighted and limits understanding of the myriad ways that life is experienced. Exploratory data mining, though the use of regression tree and random forests analysis, offers a novel approach to this research problem. Three studies, assessing predictors of perceived stress in middle and later life, are presented here using these methods. Generally speaking, regression trees can offer an idiographic perspective by highlighting complex interactions resulting in multiple routes to similar levels of stress and how individuals with comparable scores on a given predictor may report very different levels of global perceptions of stress based on their other life circumstances. Random forests, in contrast, typically offer more nomothetic information by examining many trees and providing indices of predictor importance that can direct future studies. Together, these techniques offer the opportunity to inform life stress theory, direct intervention and prevention efforts, and guide future work.
