Reliability of Time Estimation: Effect of a Preceding Reproduction Series on the Reliability of Subsequent Verbal Estimates of the Same Standard Stimuli

Abstract

32 Ss, 16 in an experimental group and 16 in a control group, gave two verbal estimates of each of 8 standard intervals ranging from 300 to 1000 msec. in 100-msec. steps. The experimental group reproduced each standard 12 times before verbally estimating it, while the control group had no prior experience with the standards to be verbally estimated. The mean reliability of the verbal estimates of the experimental group was significantly higher ( p < .02) than that of the control group. This finding was held to account for the high reliability coefficients reported for the method of verbal estimation by a number of comparative studies employing several methods on the same group of Ss. The result suggests that serial position of the measuring methods may spuriously inflate the reliability coefficients of the method of verbal estimation when that method is used subsequent to the method of reproduction.