Re: st: Reganat command in Stata Journal (2013) 13 #1

[Daljit Dhadwal <ddhadwal@gmail.com>]

I figured out why the output from the regression anatomy (reganat)
command is different than what is shown in the Stata Journal article
when using the semip option in Stata 10.1. The reganat command uses
the pcorr command to display the table of partial and semipartial
correlations. In Stata 10.1, the pcorr command only shows the partial
correlations and not the semipartial correlations and it’s also why
the model´s variance decomposition table comes up blank. It looks like
the pcorr command was updated in Stata 11 to show the semipartial
correlations.

Daljit Dhadwal

On Fri, May 31, 2013 at 1:05 AM, Daljit Dhadwal <ddhadwal@gmail.com> wrote:
> I was working through the example (on page 105) from the article
> Regression anatomy, revealed in Stata Journal Vol 13 #1 page 92-106,
> and I’m getting different output than in the article when I use the
> semip option. The partial correlation information is displayed but the
> semipartial correlation information and the model's variance
> decomposition information is missing from the table.
>
> When I issue the following commands in Stata 10.1:
>
> sysuse auto
> reganat price length mpg weight, dis(length) semip
>
> I get the following output:
>
> Regression Anatomy
> ------------------------------------------------------------------------------
> Dependent variable ...... : price
> Independent variables ... : length mpg weight
> Plotting ................ : length
> (obs=74)
> Partial correlation of price with
>     Variable |    Corr.     Sig.
> -------------+------------------
>       length |  -0.3009    0.010
>          mpg |  -0.1226    0.305
>       weight |   0.4080    0.000
>
> Model's variance decomposition                            Value          Perc.
> ------------------------------------------------------------------------------
> Variance explained by the X's individually             .              .
> Variance common to X's                                         .              .
> ------------------------------------------------------------------------------
> Variance explained by the model (R-squared)              0.3574
>
> Whereas in the article on page 105, the following output is displayed:
>
> Regression Anatomy
> ------------------------------------------------------------------------------
> Dependent variable ...... : price
> Independent variables ... : length mpg weight
> Plotting ................ : length
> (obs=74)
>
> Partial and semipartial correlations of price with
>
> Variable  Partial Corr.  Semipartial Corr.  Partial Corr.^2
> Semipartial Corr.^2  Significance Value
> length  -0.3009  -0.2530  0.0906  0.0640  0.0102
> mpg  -0.1226  -0.0991  0.0150  0.0098  0.3047
> weight  0.4080   0.3582   0.1664  0.1283  0.0004
>
> Model´s variance decomposition   Value   Perc.
> Variance explained by the X´s individually  0.2021   0.5656
> Variance common to X´s    0.1553   0.4344
> Variance explained by the model (R-squared)  0.3574
>
> So basically when I use the semip option, I’m missing the semipartial
> correlations and the model’s variance decomposition values.
>
> Any help with this would be appreciated.
>
> Thanks,
>
> Daljit Dhadwal
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/