Model Specification

Model specification refers to the process of expressing a theory in mathematical (functional) form. The choice of model specification affects the validity of causal inferences. Arguably, severe specification errors impede the validity of inferences more than the choice of a suboptimal estimator. Yet applied researchers rarely follow a strategy when trying to develop and potentially improve the specification of their model. Rather, the choice of a model specification usually depends on a crude mixture of what is common in the field of research, methodological fads, and individual intuition. Misspecification occurs when the assumptions underlying an empirical analysis deviate from the true data-generating process. Nonrandom sampling, measurement error, model uncertainty, and a lack of independence of observations rank most prominently among the sources for model misspecification. Since researchers hardly, if ever, know the true model, most models analyzed in the social sciences will, necessarily, be misspecified. The most common misspecifications result ...

Users without subscription are not able to see the full content on this title. Please, subscribe or login to access all content on this website.