ABSTRACT
Multiple regression offers analysts one of the most powerful and useful tools for quantitative analysis. With the exception of descriptive statistics, it is the most widely used quantitative method. There are three primary reasons for its popularity. First, it is accessible. Regression analysis is relatively easy to use and understand, and estimation software is widely available. Second, the multivariate linear specification is robust. Many relationships have been found empirically to be linear. The linear specification is the simplest, and thus always appropriate as a first approximation of a causal relationship. Moreover, we often do not know enough about the relationship under study to justify an alternative (nonlinear) specification. Third, the results of regression analysis have proven to be very useful, both for predicting or forecasting and for explanation (i.e., determining the causes of a phenomenon). It is the ability of multivariate regression to control for confounding influences on the relationship under study that makes it a particularly powerful tool for explanation.
