Incidental references in a number of recent papers suggest that often similar answers will be obtained to a problem of social research even though yardsticks used to measure the factors involved differ from each other. In other words, indices which might not correlate completely with each other may have practically the same correlation with a third variable. For example, Smith made a study comparing the way people actually spend their leisure time with an average estimate of the leisure-time activities given by the same people at the beginning of the study. Smith finds an average correlation of only +0.50 to +0.70 between the estimated and the actual time budget. His own summary shows, however, that if he characterized the leisure-time activities of his group of 80 subjects by a graphical profile, he would get nearly the same results whether he used the estimates or the time actually spent. The rank correlation between the two series of leisure-time data is 0.98. Jenkins points out that if housewives are interviewed twice about their last purchases, 10–15% of the women change the name of the brand from one interview to another; the total distribution of brands mentioned, however, remains the same because the individual changes cancel each other.