Power analysis in biological anthropology

Abstract

AbstractPower in statistical terms is the probability of rejecting a false null hypothesis. Power analysis, although generally not employed by biological anthropologists, is especially useful in interpreting statistical tests in which the null hypothesis is not rejected, and in determining appropriate sample sizes. In this report, the utility of statistical power analysis and the level of variation in power is illustrated by performing power analysis on paleopathology data (Cohen and Armelagos: Paleopathology at the Origins of Agriculture, 1984). The power analyses indicate that most samples are adequately large for detecting large differences among groups, but are inadequate for detecting small differences. Statistical analyses with inadequate power to detect small to medium effects may be of questionable utility in many areas of biological anthropology, as well as in paleopathology. The application of power analysis is recommended for determining necessary sample size in advance of data collection and for interpreting negative results.