Erratum

Abstract

In a recent article in this journal, I reported a failure to replicate the positive correlations Bruce Russett found between inequality of land tenure and political instability.2 Unfortunately, two of the instability indicators I used were untransformed counts (Deaths from domestic political violence, 1948–67 and 1961–65), whereas Russett had employed the logarithm of a similar measure. Taking logarithms is advisable when using counts in order to reduce skewness.3 Correlations between inequality and the logarithms (to the base ten) of the Deaths variables (plus one, as the log of zero is undefined) are greater than those I reported, though not so strong as Russett found. For 1961–65, the correlation is .29; for 1948–67, .20. The former is significant at the .05 level; the latter is not. When the Gini coefficients are weighted by the percentage of each nation's labor force in agriculture in i960,4 the correlations with the logged Deaths variables were .63 and .61—both highly significant. These figures especially support Russett's conclusions.