Happiness & Public Policy

R.J. Rummel has posted a nice summary of Inglehart and Klingermann’s excellent paper, Genes, Culture, Democracy, and Happiness [pdf].

. . . they did a multiple regression of well being against measures of a nation’s economic development, whether it was historically ruled by Protestant elites or not, its years under communist rule, and its measure of freedom. These variables account for 80 percent of the variation in well being, a remarkable fit. They then removed independent variables with low significance in stages to achieve of fit of 78 percent of the variance with three significant variables, which in the order of their significance are: GNP per capita, years under communist rule, and freedom. Aside from applying sample tests of significance to a universe of cases, a problem with their analysis, is the high multicollinearity among these three variables (on this problem, see my blog here). Without eliminating this intercorrelation, it is impossible from this regression alone to determine what variables are dominant.

The whole post’s worthwhile.