Skip to Main content Skip to Navigation
Journal articles

What to Do (and Not Do) with Multicollinearity in State Politics Research

Abstract : State politics scholars often confront data situations where the explanatory variables in a model are highly related to each other. Such multicollinearity (“MC”) makes it difficult to identify the independent effect that each of these variables has on the outcome of interest. In an effort to circumvent MC, researchers sometimes drop collinear variables from the regression model. Using simulated data, we demonstrate the implications that MC has for statistical estimation and the potential for introducing bias that the omitting-variables approach generates. We also discuss MC in the context of multiplicative interaction models, using research on the influence of the initiative on policy responsiveness as an applied example. We conclude with advice for researchers faced with MC in their datasets.
Document type :
Journal articles
Complete list of metadata
Contributor : Hélène Saint-Gal Connect in order to contact the contributor
Submitted on : Friday, April 8, 2022 - 2:53:37 PM
Last modification on : Saturday, April 9, 2022 - 3:31:07 AM

Links full text




Kevin Arceneaux, Gregory Huber. What to Do (and Not Do) with Multicollinearity in State Politics Research. State Politics and Policy Quarterly, SAGE Publications, 2007, 7 (1), pp.81-101. ⟨10.1177/153244000700700105⟩. ⟨hal-03635450⟩



Record views