Statistical sampling and hypothesis testing in orthopaedic research.

Abstract

The purpose of the current article was to review the process of hypothesis testing and statistical sampling and empower readers to critically appraise the literature. When the p value of a study lies above the alpha threshold, the results are said to be not statistically significant. It is possible, however, that real differences do exist, but the study was insufficiently powerful to detect them. In that case, the conclusion that two groups are equivalent is wrong. The probability of this mistake, the Type II error, is given by the beta statistic. The complement of beta, or 1-beta, representing the chance of avoiding a Type II error, is termed the statistical power of the study. We previously examined the statistical power and sample size in all of the studies published in 1997 in the American and British volumes of the Journal of Bone and Joint Surgery, and in Clinical Orthopaedics and Related Research. In the journals examined, only 3% of studies had adequate statistical power to detect a small effect size in this sample. In addition, a study examining only randomized control trials in these journals showed that none of 25 randomized control trials had adequate statistical power to detect a small effect size. However, beta, or power, is less well understood. Because of this, researchers and readers should be aware of the need to address issues of statistical power before a study begins and be cautious of studies that conclude that no difference exists between groups.