Re: Using resampling methods to detect influential points

Steve Samuels pointed out that this post refers to the unstandardized
dfbeta values. We compute unstandardized dfbeta values, for example,
when we use the -predict- command after -stcox-. On the other hand,
the -dfbeta- command used after -regress-, computes the standardized
dfbeta values. You can find a discussion about these values in the
section "Remarks and example for dfbeta", in the entry for
regress postestimation in the [R] Manual. Steve also has posted
a couple of answers to Statalist about computing standardized
dfbeta values using jackknife.
See
http://www.stata.com/statalist...
http://www.stata.com/statalist...