Multiple linear regression - Accounting for multiple simultaneous determinants of a continuous dependent variable

More often than not, when one's impulse is to conduct a series of separate simple regressions involving the same response variable, multiple regression should be used instead. The flexibility of multiple regression allows elegant, insightful, and often the only correct analysis. Simple nonlinearities and interaction effects can be introduced to extend the utility of this method well beyond that of simple regression. As with any multivariate statistical technique, however, it is possible to make substantial errors if the method is applied blindly without appropriate consideration of the underlying assumptions, correlations among predictors, influential observations, and thoughtful exploration of model structure. © 2008 American Heart Association, Inc.