[Gretl-users] t-stats and p-values after mle

[Allin Cottrell]

On Tue, 27 May 2008, Stefano Balietti wrote:

> Is there any accessor to catch the matrix of the p-values and 
> the matrix of the t stats after performing a ml estimation? 
> $test and $pvalues don't seem to make my case, as they return a 
> scalar and anyway they are not filled by the mle command. Maybe 
> they should be able to return a matrix as well.

Too complicated, IMO.  A case for do-it-yourself.

After mle:

 matrix stats = $coeff ./ $stderr
 stats = stats ~ (2 * pvalue(z, abs(stats)))
 colnames(stats, "t-ratio p-value")
 printf "%12.5f\n", stats

Glad you asked, because it made me realize that the use of 
pvalue() with a matrix argument was broken by a recent "fix" -- 
but this now works properly in CVS.

Allin.