🚀 go-pugleaf

From rar Wed Jan 22 09:40:59 1992 remote from sequent
Subject: Re: degrees of freedom.
To: statl@earm.mcgill1
Date: Wed, 22 Jan 92 9:40:59 WET
X-Mailer: ELM [version 2.3 PL11]
Sender: rar

The discussion about dividing by (n-1) to estimate the variance
seems to have overlooked that you can divide by (n) or (n-1). In the
first case you get THE sample variance; in the second case you get
AN unbiased estimate of the population variance, under the assumption
that the sample is a random sample from an infinite homogeneous
population.

I found to my horror a couple of years ago that in Minitab (Version 5
running on an ICL 3980 under VME), the command

let s=stdev(x)  gave one answer but
describe x      gave a different value for the SD.

In one place (n) was used, in the other (n-1). I forget which was which,
and I may have got the syntax or function names slightly wrong, but the
two uses were not highlighted by the program nor in the manual so it put
me off using Minitab. This is after all a program in worldwide use for
almost 20 years now!