RESUMEN
Strength training for older adults is increasingly common, yet surprisingly little research has evaluated the reliability of strength testing protocols in this population. Thirty-three volunteers (17 women, 16 men; 72 +/- 6 years) were tested for strength of the knee and ankle using a Biodex 3 dynamometer on 3 separate occasions. The peak torque and work for each test was analyzed for reliability over the last 2 visits using limits of agreement (LOA). The magnitude of the systematic bias was 8 Nm or less for the peak torque and 5 J or less for the work measures. The random error ranged from 9 to 20 Nm and 6 to 24 J for peak torque and work, respectively. Heteroscedasticity was present in 8 of the 20 measures. The ratio LOA ranged from 21% to 43% for these peak torque and work measures. The total error of each strength measure, which was mostly comprised of random error, can be applied to interpretation and development of training protocols for the older adult.