SPSS is the statistical software of choice in Indiana University of PA, USA. Many applications
require a regression method. In the Applied Research Lab, a statistical consulting
arm of the graduate school in our university, we often face the problem of parameter estimation
when there is multicollinearity. One simple solution to this problem is to drop some
of the highly correlated variables. This strategy usually works well. However, there are
times when the variables are too important to be excluded from the analysis. One strategy
is to apply Ridge Regression in such cases. Under some conditions, Ridge Regression has,
in theory, been shown to be effective in dealing with multicollinearity. However, it was
unclear in many applications that these conditions were satisfied. Indeed, we were
never sure if Ridge Regression provides a better model in these applications. In many
cases, we are also interested in how well the SPSS Ridge Regression procedure works.
We thus conducted this simulation study to evaluate the SPSS ridge regression.