[Gretl-devel] different treatment of d.o.f. correction between sd()/sdc() and corr()/mcorr()

Allin Cottrell cottrell at wfu.edu
Wed Jul 1 14:46:44 EDT 2015


On Wed, 1 Jul 2015, Sven Schreiber wrote:

> Am 01.07.2015 um 19:21 schrieb Allin Cottrell:
>> On Wed, 1 Jul 2015, Sven Schreiber wrote:
>> 
>> 
>> This may not provide a very compelling rationale for what we do now, but
>> the difference between sd() and sdc() is documented in the help for
>> those functions.
>> 
>
> Well only if you equate the term "sample sd" in the doc for sd() with the 
> meaning of T-1. When I read the docs (and I did!) it didn't mean anything 
> special to me, or rather I was thinking it would relate to the currently 
> active sample (as opposed to the whole workfile range). So a bit opaque IMHO.

I guess there's a national difference here: in English statistical 
parlance, "sample standard deviation" is a term of art which 
definitely implies use of T-1 in the denominator. But the wording 
could be made more explicit.

Allin


More information about the Gretl-devel mailing list