ABSTRACT
Regression is easily the most frequently used statistical technique in the social sciences (Allison, 1999), and with good reason. The regression model is both flexible and robust. It can be employed for any response variable that is at least approximately interval or even binary. The independent variables can be characterized by any level of measurement. It can easily be adapted to handle many types of nonlinear and nonadditive relationships between the response and explanatory variables. A variety of mechanisms may be employed to get around violations of model assumptions. It affords a good first approximation to analyses requiring more complex statistical techniques. Additionally, software for estimating regression models is simple enough to be included in even the most rudimentary statistical packages.
