A Meta-Analysis of Physical Activity Interventions in Higher Education Settings

Background/Purpose: Physical activity (PA) intervention programs have been implemented to promote PA among university students for years. No evidence, however, is available  to ascertain the overall effectiveness of such interventions. This study, therefore, was aimed to evaluate the effectiveness of PA interventions to increase PA among college-aged population based on literature synthesis.

Method: A systematic search strategy was used to identify relevant studies published in the English-language in health, fitness, physical education, and psychology from 2000 to 2012. The key words of “PA and exercise interventions” and “university or college students” were used to search EBSCOhost research databases and PubMed. Eligible studies were required to target university student population using PA as one of the outcome variables in higher education settings. Studies were coded independently by three investigators after an acceptable reliability was reached through an intensive practice among the investigators. Because studies on the topic used different research designs, four comparisons was used to calculate the effect size: (a)post-test PA level between control and treatment group; (b)pre- and post-test change in PA level among control groups; (c) pre- and post-test change in PA level among treatment groups; and (d) pre- and post-test change in PA between control and treatment group. Mean effect size (Hedges’s g) for each comparison was calculated based on computer software Comprehensive Meta-Analysis 2.0.

Analysis/Results: A total of 18 articles were identified for meta-analysis. The mean effect size of post-test difference between control and treatment group was relatively small yet significant (g = .33, p< .01). Similarly, among treatment groups, the mean effect size of pre- and post-test difference was significant (g = .24, p< .01). The mean effect size of pre- and post-test PA level change between treatment and control group was near moderate level (g = .42, p< .01). Among control groups, the mean effect size of PA level change between pre- and post-test was small and insignificant (g = .08, p> .05). 

Conclusions: Results of the meta-analysis support the overall success of PA interventions among college-age population. However, none of the effect size in this study reached moderate level (i.e., g ≥ .50), indicating that the effectiveness PA intervention program in college settings is still in need of further improvement. Meanwhile, the ability of PA interventions to produce meaningful change in long-term effects remains unanswered as few studies on the topic have been reported. It was also recommended that the potential moderators in PA intervention effectiveness should be investigated in future studies.