I am working on a multivariate regression model that examines the relationship between an index of state consolidation as the dependent variable and a number of predictor variables. I have conducted some basic tests of the predictor variables and transformed them, etc.
However, the correlation matrix and other tests seem to show multicollinearity between some of the predictors. Theoretically, I do not see how they are related to each other. Accordingly, I tried to center and standardize the predictors. I used the commands beta and another called stdbeta. I also created some modified predictors the old-fashioned way subtracting from the mean and dividing by the standard deviation. The stdbeta command did not work and the multicollinearity persisted using the modified predictors.
I am at an impasse and work greatly appreciate some insight into this.
I have attached the do file.
Thank you,
J. David Granger
However, the correlation matrix and other tests seem to show multicollinearity between some of the predictors. Theoretically, I do not see how they are related to each other. Accordingly, I tried to center and standardize the predictors. I used the commands beta and another called stdbeta. I also created some modified predictors the old-fashioned way subtracting from the mean and dividing by the standard deviation. The stdbeta command did not work and the multicollinearity persisted using the modified predictors.
I am at an impasse and work greatly appreciate some insight into this.
I have attached the do file.
Thank you,
J. David Granger
Comment