Testing Whether Regression Coefficients Are Correlated

Occasionally, in a multiple linear regression model using an empirical equation, two or more independent variables are correlated among themselves, that is, one independent variable is a linear function of another independent variable. Intercorrelated variables are said to exhibit multicollinearity. A nonchemical example of multicollinearity might exist in the relationship between 

If two independent variables are perfectly correlated (i.e., one is an exact linear function of the other) then LINEST will return NUM in all cells. If they are only approximately correlated, then LINEST will return regression parameters that can be very misleading R - can be very close to 1, but the coefficients may have no meaning. The standard errors of one or more of the coefficients will probably be fairly large. 

---