The Survey Research About The Status Of Experimental Design And Corresponding Statistical Analysis Methods In Sports Science

Good experimental design can help us define appropriate sample size, strict control experimental error, so that the experimental effect can be highlighted, with less manpower, material and time to obtain a more satisfactory results, and answer the hypothetical questions. This study surveyed the situation of the application about experiment design and corresponding statistical analysis methods, which was from 611 theses of the three core journals right from 2000 to 2008. The main results and conclusions are:Results:1. From 2000 to 2008, the application ratio experimental design methods stay unchanged, showing three levels. The higher application ratio to lower application ratio was completely random design(53.8%), single group design, factorial design(all about 20%), randomized block design and other design(all less than 5%).From 2000 to 2008, the application ratio of t test had a drop trend; the application ratio of single-factor analysis of variance, complex statistical analysis methods (including multi-factor analysis of variance, repeated measure analysis of variance, etc.) was on the rise; the application ratio of statistical methods is not known, only have P value had a drop trend.2. From 2000 to 2008, the reasonable ratio of the experimental designs had not changed. The reasonable ratios of single-group design, completely random design, randomized block design were relatively higher, and the reasonable ratios of factorial design and other designs were lower. From 2000 to 2008, the reasonable ratio of the statistical analysis was increased; the unreasonable ratio remained unchanged; not clear ratio had a downward trend. 3. From 2000 to 2008, the frequency of application of experimental design principles has not changed; there was no significant change in the state of "no appliance test of statistics method ", and the frequency was 96.5%. From 2000 to 2008, there was no significant different in the misuse rate of t test, the average misuse rate was 76.1%; non-single-group design for experimental research papers,83.7% did not pre-test the sub-combination of rational proof.Conclusions:There were some situations of experimental designs misuse and against principles of experimental designs; from 2000 to 2008, there was improved trend of statistics analysis, but still not optimistic.