Re: weighted variance comparisons

"John Kane" <jrkrideau@xxxxxxxxx> wrote in message
news:1188314968.618827.325290@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
...
To follow up peterp's suggestion of alternative software: R is Open
Source (hence free to the user) and can be downloaded from
http://www.r-project.org/.

It is not really safe to do stats with a spread***.

I have to strongly object to the last remark, which I found utterly déplacé
(i.e., completely gratuitous - as the vast majority would have put it, but I
chose not to merely to emphasise the "European intellectual" type of
fussiness of my post).

Namely, I fully agree with the sound, polite and kind peterp's arguement
("If you will keep dealing with the combination statistics X large
datasets, consider investing some time in learning the basics of the R
statistical programming language"), combined with objective and benevolent
assessment of spreadsheets ("... your spreadsheet program provides variance
function for frequency-summarized data, I noticed that neither excel and
openoffice calc do").

And I admire anyone capable of using R, and I most strongly encourage
everyone I can to learn R, be it students or researchers, be it in
statistics or psychology or physics or elsewhere.

And I've had the privilege of personally meeting prof. Harrell, and there's
really no need for my praise of his magnificent work here, be it his S/R
libraries or other software or publications.

And mentally limited middle-aged non-mathematician as I am, I happen to be
the co-author of the first R package accepted to CRAN from my country
(which, tiny as it is, is not at all short of heavyweight statisticians).

However, I also dare claiming to be at least a bit of an Excel expert, and
quite widely if not even profoundly knowledgable on the topic of
spreadsheets in statistics with it's myriad of pros and con(tra)s (how about
that for many meanings and puns?) and aspects. I even wrote a paper related
to those matters (it's publicly accessible and easily found and retreived by
anyone interested), which containins more than plenty of links and
references that one should - at least in my opinion - study in order to be
able to make qualified judgements on the topic. But giving the link to the
paper would be the very same kind of marketing-style buzz-word-like shallow
pseudo-arguement that I see in the statement which provoked my response.

To summarise for the original poster: do learn R, by all and any and every
means! But if you also like or even fall in love with (go crazy for, feel
profoundly intrigued by, etc.) spreadsheets, or anything or anyone else,
follow your passion, go ahead, just do it, be neither a conformist nor an
anti-conformist (beacuse it's the same thing), be a non-conformist, study
and learn and explore everything in this world all the time!

Because some "common wisdom" (deplorable as it mainly is) actually holds
(whether by virtue of type I or type II error, my dear Watson, is irrelevant
here :), namely that there is no other world than this one, and that nothing
is safe.

Best regards,
Gaj Vidmar


.

Quantcast