Evidence for Reliable Individual Differences in Intra-Individual Variability

Abstract

In reliability theory, the deviations in an individual's performance from his own mean score are referred to as error and the assumption is made that error scores in a series of repeated tests will be uncorrected. To test this assumption, 24 college males were given 175 RT trials a day for 5 days (the first day served as a warm-up session and was not used in the analyses). Within each test day, a short rest period was given following every block of 35 trials. When data were examined on a day-to-day basis, i.e., the total 175 trials per day were used, the reliability coefficients were high (.811 to .866); on a block-to-block within-days basis, i.e., the total 175 trials per day were subdivided into 5 blocks of 35 trials, the coefficients were lower but statistically significant (the average block-to-block correlations for test Days 1, 2, 3 and 4 were .639, .704, .734 and .646 respectively). Re-evaluation of reliability theory and its assumption that an individual's deviations from his own mean ability represent error is needed. Present data suggest these deviations from mean ability in a motor skill actually reflect a biological variability and, as such, should be referred to as intra-individual variability.