Re: [Gretl-users] Corrgm function (after thinking)
On Tue, 8 Sep 2009, Ofer Cornfeld wrote:
Thanks. I have attached a small script output with the BUG
displayed. You can see the first corrgm is executed with no
problem, but the second one fails with the error message. Hope
this will help.
Yes. The short answer is, I'm sorry to say, that the corrgm()
function is nastily broken for datasets including missing values
(for the case where a third argument is given, at any rate) and
should not be trusted at all in that context! I'll put out a
proper fix as soon as I can.
Meanwhile, please note that the xcorrgm command produces correct
results for the cross-correlogram, although of course it doesn't
give you the option of assigning the result to a matrix.
Allin Cottrell