Reliability assessment of pedometer activity measurements

Pedometers (step counters) are inexpensive devices for obtaining naturalistic activity measurements. The existing reliability literature on pedometers is inconsistent. This article presents reasons for the disparity including (a) distinguishing between instrument reliability (Experiment 1) and clinical repeatability (Experiments 2–4) and (b) discussing methods of evaluating the reliability of activity measuring instruments. Experiment 1 shows that pedometer measurements have an uncertainty of approximately 5% (the test-retest correlational equivalent of .97) under laboratory conditions. Experiment 2 shows that pedometer measurements in college students walking a measured half-mile have approximately the same uncertainty. Experiment 3 shows that pedometer measurements from normal children walking a measured half-mile have an uncertainty of approximately 18% (the correlational equivalent of .91), while measurements from mildly hyperactive children have an uncertainty of approximately 29% (the correlational equivalent of .84). Experiment 4 shows that electronic step counters have an uncertainty of approximately 9% in normal and mildly hyperactive children and about 12% in clinically hyperactive children.