Logistic Regressions: Habitual And Regular Exercisers And Perceived Barriers To Being Active

Abstract

PURPOSE: This study examined (a) what perceived physical activity (PA) barriers would significantly prevent people from becoming habitual and regular exercisers (HRE) and (b) how accurately the PA-barrier predictors would predict HRE and non-HRE. METHODS: Exercise Stages of Change survey (Marcus & Forsyth, 2009) and Barriers to Being Active Quiz (CDC, 1999) were administered to 286 university students (mean age: 20.98 ± 1.49; 117 females) in the US. The former classified participants into two groups of exercisers: 167 students (60 females) in HRE, performing regular exercise as a habit and meeting Federal PA Guidelines (FPAG) for at least six months; and 119 (57 females) in non-HRE, including those having no exercise, those having some exercise, and those having regular exercise but meeting FPAG for fewer than six months. The barrier quiz produced seven PA-barrier variables: lack of time, social influence, lack of energy, lack of willpower, fear of injury, lack of skill, and lack of resources. Each variable had a score ranged from 0 to 9, a high score indicating high barrier. A logistic regression was run with the seven PA-barrier variables and sex as predictors, and HRE as the predicted event. RESULTS: The results showed the model was significant in predicting HRE vs. non-HRE with -2 log likelihood = 291.45, omnibus χ2(8) = 96.93, p < .000; and Nagelkerke R2 = .387, explaining 38.7% of variance of the outcome. “Lack of willpower” was the first significant predictor with B = -.42, Wald χ2(1) = 25.05, p < .001, and odds ratio = .66; and “lack of resources” was the second significant predictor with B = -.21, Wald χ2(1) = 3.98, p < .05, and odds ratio = .81. In addition, the correct classification rate was 84.4% for HRE, 63.9% for non-HRE, and 75.9% for the overall. CONCLUSIONS: While the predictors can explain 38.7% of variance of the outcome, only two predictors are significant in preventing people from becoming HRE while holding the other predictors constant. One-point increase in “lack of willpower” and in “lack of resources” will decrease the likelihood in the HRE by 34% (1 - .66 × 100%) and by 19% (1 - .81 × 100%), respectively. In addition, the predictors demonstrate a relatively high classification accuracy (75.9%) in HRE and non-HRE. All these results, however, are limited to university students.