Published online by Cambridge University Press: 04 January 2017
Social scientists often employ measures that tap correlated factors. This article explores the effects of these complex measures on OLS estimators and specifies some of the conditions under which researchers should or should not include them in a regression.
The derivations allow us to shed some new light on the debate about sociotropic and pocketbook voting. In a classic piece, Kramer (1983) criticized inferences about economic interests that were based on cross-sectional data. He also argued that aggregate data could be used to assess the net effects of both sociotropic interest and self-interest on the vote. In this article, by using a complex judgment measure of sociotropic and self-interested economic perceptions and a redundant measure of self-interested perceptions, we are able to produce a consistent estimate of the sociotropic effect with aggregate data. An estimate of the self-interest effect is also available, though it may not have any desirable asymptotic properties.