Help with test for dummy significance
Posted: Sat Feb 25, 2012 7:25 am
Hello,
I am fairly new to econometrics and have an assignment that I would like some clarification with.
My regression involves regressing wage on various variables including dummy variables for white, black and asian. I have run separate regressions using subsamples for specific races so far.
I am then asked to suggest a model that allows one to test the nul hypothesis of coefficient constancy across the three subsamples and carry out
the test using the appropriate F statistic.
I think this is the correct strategy, however I would greatly appreciate if someone could clarify this:
1. Run the regression of y on all of the independent variables and 2 of the races (to avoid dummy variable trap)
2. Run the regression of y on all of the indepedent variables apart from the races.
Do an F test for joint restrictions using SSRr, SSRur or R^2 restricted and R^2 unrestricted.
I would greatly appreciate if someone could let me know if this is the correct method!
Thanks
I am fairly new to econometrics and have an assignment that I would like some clarification with.
My regression involves regressing wage on various variables including dummy variables for white, black and asian. I have run separate regressions using subsamples for specific races so far.
I am then asked to suggest a model that allows one to test the nul hypothesis of coefficient constancy across the three subsamples and carry out
the test using the appropriate F statistic.
I think this is the correct strategy, however I would greatly appreciate if someone could clarify this:
1. Run the regression of y on all of the independent variables and 2 of the races (to avoid dummy variable trap)
2. Run the regression of y on all of the indepedent variables apart from the races.
Do an F test for joint restrictions using SSRr, SSRur or R^2 restricted and R^2 unrestricted.
I would greatly appreciate if someone could let me know if this is the correct method!
Thanks