Findley T W
Kessler Institute for Rehabilitation, West Orange, NJ 07052.
Am J Phys Med Rehabil. 1990 Aug;69(4):209-18. doi: 10.1097/00002060-199008000-00009.
The primary statistical analysis is approached from the standpoint of what is required to publish in medical research journals. Descriptive and bivariate statistics cover the majority of medical research articles now published. Statistical guidelines for review of manuscripts are used to develop guidelines for analysis, including specifying the objective, the source of subjects and response rate, differences detectable with the expected sample size, appropriateness of statistics for one and two variables, method of presentation of results and conclusions and calculation of confidence intervals. Common mistakes to avoid include use of standard error of the mean instead of standard deviation, use of standard deviation with skewed data, failure to describe the statistical test used, multiple comparisons and failure to use special forms of t test and chi 2.
主要的统计分析是从医学研究期刊发表所需的角度进行的。描述性统计和双变量统计涵盖了目前发表的大多数医学研究文章。用于稿件评审的统计指南被用来制定分析指南，包括明确目标、受试者来源和应答率、预期样本量可检测到的差异、单变量和双变量统计的适用性、结果和结论的呈现方法以及置信区间的计算。要避免的常见错误包括使用均值标准误而非标准差、对偏态数据使用标准差、未描述所使用的统计检验、多重比较以及未使用t检验和卡方检验的特殊形式。
