Repeatability and interobserver error of digit ratio (2D:4D) measurements made by experts

Abstract

The second-to-fourth digit ratio (2D:4D) is sexually differentiated (lower in men than in women), a likely biomarker for organizational (permanent) effects of prenatal testosterone, and a correlate of many sex-dependent, hormonally influenced traits and phenotypes. The extent of 2D:4D measurement repeatability across different research groups is unknown. This study assessed the repeatability and interobserver error of 2D:4D measurements made by 17 experts (researchers who have contributed to the 2D:4D literature). Results indicate that 2D:4D, because it is a ratio variable, is notably less precisely measurable than finger length. Absolute-agreement intraclass correlation coefficients for these traits are about 0.75 vs. about 0.95, respectively. Associations of 2D:4D with target traits were usually of small size; measurement unreliability attenuates effects. This may explain some null findings and replication failures encountered in 2D:4D research. However, agreement levels are still sufficient to permit conclusions about findings from different research groups. One implication of this is that the marked geographical and population differences in typical 2D:4D levels, as reported in the literature, are veridical, and not due to divergent measurement habits across research groups. Some practical recommendations for 2D:4D measurement are offered. Am. J. Hum. Biol. 19:142–146, 2007. © 2006 Wiley-Liss, Inc.