« Hypothesis Testing | Main | Regression Theory »
September 15, 2005
Ordinary Least Squares
Issue in choosing the dependent and independent variable:
We also talked about the difference between an accounting identify (see above) and a behavioral relationship.
Unlike covariance and correlation, ordinary least squares assumes a particular orientation for the deviations from a straight line. This leads to some interesting differences in formulas.
Why isn't the R-squared the best way to pick the right equation?
Thinking ahead to inference:
Posted by bparke at September 15, 2005 09:33 PM