|（School of Physical Education，Chuzhou University，Chuzhou 239000，China)
Abstract: Based on the data of Chinese General Social Survey 2012 (CGSS2012), by applying multi-model identifica-tion technology in which the Heckman selection model was combined with the instrumental variable model, the author corrected the problem of endogenous bias brought by such causes as sample selection and variable omission, and veri-fied the causal mechanism between sports behavior and health return. The author also estimated the net effect of health return of sport. The author revealed the following findings: 1) between sports behavior and health return there is not only unidirectional causal correlation (i.e. sport promotes health), but also a bidirectional functioning mechanism; 2) the effect of sports resource distribution on people’s sports participation is not homogeneous, its degrees of functioning on different groups are different; 3) the unbalance of sports resource layout will trigger problems such as sports partici-pation inequality and health return inequality etc.
Key words: sports metrology；multi-model identification；sports behavior；health return；causal effect