Analysis of Paired Data in Physical Therapy Research: Time to Stop Double-Dipping?

One of the most fundamental requirements of many statistical analysis procedures is that each data point must represent an independent observation. However, researchers in the field of musculoskeletal therapy are often faced with the dilemma of how to manage measurements obtained from both limbs of the same participant. In this situation, do we have a sample size of 30 participants, or 60 limbs? Similar problems are encountered in podiatry, where it is not uncommon for researchers to measure both feet, and ophthalmology, where measurements are frequently performed on both eyes. Failing to address the conceptual and statistical issues associated with paired data is common in musculoskeletal research and has the potential to seriously distort research findings. J Orthop Sports Phys Ther 2005; 35(8):477–478. doi:10.2519/jospt.2005.0108