Experimental evidence for the intention-behavior relationship in the physical activity domain: a meta-analysis.

OBJECTIVE: Most contemporary theories of physical activity include an intention construct as the proximal determinant of behavior. Support of this premise has been found through correlational research. The purpose of this paper was to appraise the experimental evidence for the intention-behavior relationship through meta-analysis. METHODS: Studies were eligible if they included: (1) random assignment of participants to intervention/no intervention groups; (2) an intervention that produced a significant difference in intention between groups; and (3) a measure of behavior was taken after the intention measure. Literature searches were concluded in December 2010 among five key search engines. RESULTS: This search yielded a total of 1,033 potentially relevant records; of these, 11 passed the eligibility criteria. Random effects meta-analysis procedures with correction for sampling bias were employed in the analysis. The sample-weighted average effect size derived from these studies was d+ = .45 (95% CI .30 to .60) for intention, yet d+ = .15 (95% CI .06 to .23) for behavior. CONCLUSIONS: These results demonstrate a weak relationship between intention and behavior that may be below meaningful/practical value. We suggest that prior evidence was probably biased by the limits of correlation coefficients in passive designs. It is recommended that contemporary research apply models featuring intention-behavior mediators or action control variables in order to account for this intention-behavior gap.