Failure to find statistical significance in left-handedness and pathology studies: A forgotten consideration
Failure to find statistical significance in left-handedness and pathology studies: A forgotten consideration
复制标题
未能在左撇子和病理学研究中找到统计意义：一个被遗忘的考虑因素
DOI：
10.3758/bf03334957
复制
发表时间：
1993
期刊：
影响因子：
--
通讯作者：
S. Coren
中科院分区：
文献类型：
--
作者：
S. Coren
在目前关于左撇子作为各种病理条件标志的争论中，许多已发表的研究未能复制其他研究者报告的左撇子与各种危险因素之间的关联。这篇论文表明，许多这样的研究根本缺乏统计能力来做到这一点。如这里所示，问题通常是由条件的样本量不足组成的。例如，给定1000个对象的样本，右撇子的真实病理率为10%，即使左撇子的风险比右撇子高50%（真实病理率为15%），统计能力也只有3。这意味着在70%的研究中，这种差异在p < 0.05的情况下不会达到统计学意义！一个数字允许估计统计功率在典型条件下与惯用手性研究提供。
In the current controversy over left-handedness as a marker for various pathological conditions, a number of published studies have failed to replicate associations between handedness and various risk factors reported by other investigators. This paper demonstrates that many of these studies have simply lacked the statistical power to do so. As demonstrated here, the problem usually consists of inadequate sample sizes for the conditions. For example, given a sample of 1,000 subjects and a true rate of pathology of 10% in right-handers, even if left-handers have a 50% higher risk (true rate of pathology, 15%) the statistical power is only.3. This means that this difference would not reach statistical significance at p < .05 in 70% of the studies with a sample of this size! A figure allowing estimation of statistical power under typical conditions associated with handedness research is provided.