Health Performance Within the Campbell Paradigm: irt Models for Testing new Approaches in Health Psychology

Abstract

In health psychology, behaviors such as screening for cancer, calories counting or fastening seatbelts have been traditionally considered independent. Behaviors’ independence is commonly judged on the basis of inter-correlations. If we, however, conceptualize health behaviors as steps of different difficulty towards achieving a particular goal (i.e., being healthy) health behaviors appear interdependent. This claim, coined the Campbell paradigm, was verified in two studies, with samples of Dutch (N = 391) and Polish adults (N = 436). A one-parameter logistic Rasch model within IRT was applied to corroborate one-dimensionality of a comprehensive health performance measure composed of behavioral self-reports. Self-reported behaviors associated with different domains such as sustenance, hygiene, and physical exercise formed a homogenous class. A more complex five-dimensional model did not predict the data significantly better than a parsimonious one-dimensional version. Predictive, divergent and convergent validity of the health performance measure was verified in the context of health-behavior relevant variables, personality traits and objective indicators of health such as body composition. Our findings speak of the psychological and formal unity of health performance that would be hard to establish with standard methods within the Classical Test Theory.