Abstract: Traditional statistics methods and meta-analysis were compared in both experi-mental and overall control groups in our experimental study using computerconstruction. Under the condition of 0.50 standard diviation higher in the overall ex-perimental group than in the overall control group, the satistics testing power of thetraditional t testing was only 41%; while using meta-analysis method to analyse sta-tistics, the result was highly consistent with computer construction model. Thus it wassuggested that meta-analysis method had its validity and reliagility in psychologicalexperimental study and proving theories as well as the reality of a challenge thattraditional statistics method faces.

Key words: meta- analysis, effect size, jpower