Temporal Focus and Random Error in Cross-Cultural Hypothesis Tests

Abstract

Murdock and White (1969) have argued that in cross-cultural hypothesis tests where the variables measured are not focused (pinpointed in space and time to a single local community), we cannot assume functional relation ships between those variables. Since most of the published cross-cultural studies do not report focusing, Murdock and White cast doubt on the find ings of a few hundred studies. I argue that focusing should be used to increase data quality, but that a lack of focus only increases the amount of random measurement error, which tends to lower correlations. Therefore, if focusing is not applied and a significant correlation is found, it means that the actual association is probably even higher than the one found. This was tested by coding two variables, once with and once without a temporal focus. Both correlations were statistically significant, but the time- focused correlation was 23 percent higher. The test was then repeated on a second sample, drawn from a different universe, and similar results were found. In this instance, the time-focused correlation was 13 percent higher.