de Groot Jasper H B, Smeets Monique A M
Department of Social, Health, & Organizational Psychology, Utrecht University, PO Box 80140, 3508 TC Utrecht, The Netherlands.
Chem Senses. 2017 Oct 1;42(8):663-673. doi: 10.1093/chemse/bjx049.
Alarm pheromones are widely used in the animal kingdom. Notably, there are 26 published studies (N = 1652) highlighting a human capacity to communicate fear, stress, and anxiety via body odor from one person (66% males) to another (69% females). The question is whether the findings of this literature reflect a true effect, and what the average effect size is. These questions were answered by combining traditional meta-analysis with novel meta-analytical tools, p-curve analysis and p-uniform-techniques that could indicate whether findings are likely to reflect a true effect based on the distribution of P-values. A traditional random-effects meta-analysis yielded a small-to-moderate effect size (Hedges' g: 0.36, 95% CI: 0.31-0.41), p-curve analysis showed evidence diagnostic of a true effect (ps < 0.0001), and there was no evidence for publication bias. This meta-analysis did not assess the internal validity of the current studies; yet, the combined results illustrate the statistical robustness of a field in human olfaction dealing with the human capacity to communicate certain emotions (fear, stress, anxiety) via body odor.
警报信息素在动物界被广泛使用。值得注意的是，有26项已发表的研究（N = 1652）强调了人类具有通过体臭将恐惧、压力和焦虑从一个人（66%为男性）传递给另一个人（69%为女性）的能力。问题在于该文献的研究结果是否反映了真实效应，以及平均效应大小是多少。通过将传统的荟萃分析与新颖的荟萃分析工具、p曲线分析和p均匀技术相结合来回答这些问题，这些工具可以根据P值的分布表明研究结果是否可能反映真实效应。传统的随机效应荟萃分析得出了小到中等的效应大小（Hedges' g：0.36，95%置信区间：0.31 - 0.41），p曲线分析显示有证据诊断为真实效应（p < 0.0001），并且没有发表偏倚的证据。该荟萃分析未评估当前研究的内部有效性；然而，综合结果说明了人类嗅觉领域中一个研究方向的统计稳健性，该方向涉及人类通过体臭传递某些情绪（恐惧、压力、焦虑）的能力。
