Assumptions, Limitations, Practical Considerations - Assumption of Linearity
First of all, as is evident in the name multiple linear regression, it is assumed that the relationship between variables is linear. In practice, this assumption can virtually never be confirmed; fortunately, multiple regression procedures are not greatly affected by minor deviations from this assumption. However, as a rule it is prudent to always look at a bivariate scatterplot of the variables of interest. In Multiple Regression, these plots are readily available from anywhere within the program simply by requesting the spreadsheet with the correlation matrix, and then displaying the customized graph (scatterplot) by right-clicking on the desired cell and selecting Graphs of Input Data - Scatterplot from the shortcut menu. If curvature in the relationships is evident, you may consider either transforming the variables (via STATISTICA Visual Basic or spreadsheet formulas), or explicitly allowing for nonlinear components. Use Fixed Nonlinear Regression to fit various nonlinear components, that is, to test explicitly for the significance of a nonlinear component in the relationship between two or more variables (other nonlinear regression options are available in the Nonlinear Regression module).