[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: DFBETA for constant
--Roger Harbord <Roger.Harbord@bristol.ac.uk> asked early this week why
-dfbeta or -predict, dfbeta()- does not work for the constant term.
After exchanging emails with Roger, I privately sent him a do-file to
calculate the DFBETAs for the constant term based on a draft by Bobby
Gutierrez. Now I converted the do-file to an revised version of -dfbeta-,
called -dfbeta2-. If anyone is interested, it can be downloaded by:
. net from http://www.stata.com/users/wguan/
The syntax is basically the same as -dfbeta-, where you can use:
. dfbeta2 [varlist]
to obtain the DFBETAs for the variables specified. I also added two options:
1) -double- requests that the calculated data will be stored as double
precision. The default is float.
2) -constant- requests the DFBETAs for the constant term to be calculated.
. dfbeta2, constant /* for the constant term only */
. dfbeta2 weight, constant /* for weight and constant term */
. dfbeta2 /* for every variable in previous model
including the constant term */
Weihua Guan <email@example.com>
* For searches and help try: