Is 'exercise addiction' a problematic behavior indeed? The importance of controlling for predictors

Exercise addiction (EA) is considered a behavioral addictions. However, like most behavioral addictions, currently has no clinical diagnostic criteria. The panel evaluating its inclusion in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) concluded that there is insufficient evidence for establishing a distinct class of dysfunction for EA. There is a valid reason for this verdict because, based on the 'interactional model' (Egorov & Szabo, 2013), all cases of EA have different etiologies, and research is based on questionnaire data assumed to mirror the risk of EA (REA). However, REA may never turn into dysfunction. Furthermore, the REA shares substantial variance with passion, perfectionism, and sports commitment. Still, the extent to which these predictors or their domains 'jointly' determine the REA is unknown. Therefore, this study aimed to address this issue, based on Szabo and Kovacsik (2019), who showed that controlling for passion while testing REA differences between high and low-volume exercisers canceled out the group differences emerging without covariates. Thus, after spotting the predictors of the REA, we tested three group differences (gender, competition status, and individual vs. team sports), with and without including the predictors in the tests. We examined 1,003 regular exercisers (46.86% males) who completed validated instruments to assess the REA, its predictors, and personal exercise characteristics. While all measures showed medium to strong effect size-based correlations with the REA, a bootstrapped multiple hierarchical regression yielded only six predictors (exercise volume and intensity, harmonious and obsessive passion, rigid perfectionism, and constrained commitment), sharing 42.7% of the variance with the REA. Three bootstrapped univariate tests yielded gender, competition status, and sports-form differences in REA when the predictors were not included in the model as covariates. However, all group differences vanished when the predictors were controlled. Our findings spotlight a conceptual oblivion between what is currently conceptualized as REA and its predictors. More predictors identified in future research could entirely obscure the impact of the REA. Therefore, future research should develop more specific tools to assess EA as a potential morbidity because what we assess now may be far from dysfunctional form of exercise.
References
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: Author.
Egorov, A. Y., & Szabo, A. (2013). The exercise paradox: An interactional model for a clearer conceptualization of exercise addiction. Journal of Behavioral Addictions, 2(4), 199–208. https://doi.org/10.1556/jba.2.2013.4.2
Szabo, A., & Kovacsik, R. (2019). When Passion Appears, Exercise Addiction Disappears. Swiss Journal of Psychology, 78(3–4), 137–142. https://doi.org/10.1024/1421-0185/a000228