| Objective: Medical dissertation mainly reports the results of scientific research in medicine and clinical work summary, it is the main carrier from which many medical researchers and clinical workers get medical information. A high quality academic dissertation not only must have innovation, but also make sure its scientific, so that men can recognize the objective law of every medical phenomenon in medicine correctly, and improve the level of preventing and curing disease. But many medical researchers and clinical workers are short of knowledge which can make the use of statistical methods properly, readers and editors don't pay enough attention to these problems, which make a lot of statistical mistakes in medical dissertations and cannot reflect the results of scientific research, even make wrong results. From 70 years foreign researchers analyzed the applications of experimental design and statistical analysis method, but they didn't include sampling method and sampling size. There are reports from 90 years in our country, but they are only the investigation and analysis reports of statistical method applications, and the ranges of investigation are narrow, the analysis methods are very simple. So we choose some medical journals includingchina journal national journals^ medical university journals and province medical journals, and analyze the statistical contents of clinical and basic research dissertations in these journals, in order to know the applications and problems of statistical method in medical research, and give answers of proper scientific design > rational sampling and statistical analysis and scientific sampling size which medical research workers can refer to and use. Our propose is to cause the attentions of medical researchers, improve the quality of relative people and medical dissertations, and improve the level of scientific research.Methods: Using simple random sampling method, random choose 1012 dissertations from clinical and basic research articles , which are published on china journals > national journal medical university journals or province medical journals in 2000 year and were stored in Hebei province library and Hebei medical university library. Investigate 4 variables: experimental design(Xl)s sampling method(X2)> statistical analysis method(X3) and sampling size(X4). According to its condition divide every variable into different ranks, then give every rank different value which is regarded as observed value from best to worst. Use Q type hierarchical clustering method to cluster 1012 cases, the clustering statistic is Euclidean distance, the clustering method is furthest neighbor. Enumeration data use x 2 test to test whether the proportions are same, if the different of proportions issignificant then use x2 subdividing test. Ranked data use Ridit analysis to test whether the rank distributions are same, if the different of rank distribution is significant then use rank transformation test to multiple compare. All data analysis are use SPSS^ STATA and POMS statistical analysis software.Results: Except sampling method the rank distribution of all investigated variables are different, each x 2 value is 32.8819, 3.7832, 82.2995, 8.5823; P value is O.0001, 0.2858 , O.0001 , 0.0354. The rank distributions of experimental design in china journals x national journals and medical university journals are same, and different between province medical journals and other journals; the rank distributions of statistical analysis method between china journals and medical university journals > between medical university journals and national journals are same, and different between each other journals; the rank distributions of sampling size between china journals and medical university journals N between national journals and province medical journals are same, and different between each other journals. All 1012 cases cluster 5 type, each type content 102 cases > 357cases> HScases^ 290cases and 145 cases. Through Ridit analysis and rank transformation test the rank dist... |
PDF Full Text Request