Header

UZH-Logo

Maintenance Infos

How to use and interpret hormone ratios


Sollberger, Silja; Ehlert, Ulrike (2016). How to use and interpret hormone ratios. Psychoneuroendocrinology, 63:385-397.

Abstract

Hormone ratios have become increasingly popular throughout the neuroendocrine literature since they offer a straightforward way to simultaneously analyze the effects of two interdependent hormones. However, the analysis of ratios is associated with statistical and interpretational concerns which have not been sufficiently considered in the context of endocrine research. The aim of this article, therefore, is to demonstrate and discuss these issues, and to suggest suitable ways to address them. In a first step, we use exemplary testosterone and cortisol data to illustrate that one major concern of ratios lies in their distribution and inherent asymmetry. As a consequence, results of parametric statistical analyses are affected by the ultimately arbitrary decision of which way around the ratio is computed (i.e., A/B or B/A). We suggest the use of non-parametric methods as well as the log-transformation of hormone ratios as appropriate methods to deal with these statistical problems. However, in a second step, we also discuss the complicated interpretation of ratios, and propose moderation analysis as an alternative and oftentimes more insightful approach to ratio analysis. In conclusion, we suggest that researchers carefully consider which statistical approach is best suited to investigate reciprocal hormone effects. With regard to the hormone ratio method, further research is needed to specify what exactly this index reflects on the biological level and in which cases it is a meaningful variable to analyze.

Abstract

Hormone ratios have become increasingly popular throughout the neuroendocrine literature since they offer a straightforward way to simultaneously analyze the effects of two interdependent hormones. However, the analysis of ratios is associated with statistical and interpretational concerns which have not been sufficiently considered in the context of endocrine research. The aim of this article, therefore, is to demonstrate and discuss these issues, and to suggest suitable ways to address them. In a first step, we use exemplary testosterone and cortisol data to illustrate that one major concern of ratios lies in their distribution and inherent asymmetry. As a consequence, results of parametric statistical analyses are affected by the ultimately arbitrary decision of which way around the ratio is computed (i.e., A/B or B/A). We suggest the use of non-parametric methods as well as the log-transformation of hormone ratios as appropriate methods to deal with these statistical problems. However, in a second step, we also discuss the complicated interpretation of ratios, and propose moderation analysis as an alternative and oftentimes more insightful approach to ratio analysis. In conclusion, we suggest that researchers carefully consider which statistical approach is best suited to investigate reciprocal hormone effects. With regard to the hormone ratio method, further research is needed to specify what exactly this index reflects on the biological level and in which cases it is a meaningful variable to analyze.

Statistics

Citations

16 citations in Web of Science®
12 citations in Scopus®
Google Scholar™

Altmetrics

Additional indexing

Item Type:Journal Article, refereed, original work
Communities & Collections:06 Faculty of Arts > Institute of Psychology
Dewey Decimal Classification:150 Psychology
Uncontrolled Keywords:DoktoratPsych Erstautor
Language:English
Date:2016
Deposited On:01 Dec 2015 12:55
Last Modified:05 Apr 2016 19:36
Publisher:Elsevier
ISSN:0306-4530
Publisher DOI:https://doi.org/10.1016/j.psyneuen.2015.09.031
PubMed ID:26521052

Download

Full text not available from this repository.
View at publisher