Interaction effects in cross-national studies

Abstract

One may analyse interaction effects either by adding a multiplicative term to an ordinary regression equation or by group comparisons. The effects of human capital formation and government revenues on economic growth in LDCs serve as an example. The proposition under discussion claims that high government revenues (as a share of GDP) cancel the otherwise positive effects of human capital formation on economic growth. This proposition cannot be supported in regression analyses with an interaction term, although it receives rather strong support from comparisons of linear-additive regressions of growth of rates on human capital formation separately done for high and low revenue countries. This result leads to the question which technique and thereby which finding should be adopted. Since regression analyses with interaction terms investigate, whether an interaction of a restricted, symmetrical type applies, group comparisons should often be preferred.