[Theforum] survey update?

[Luther, Ron]

Hi Gang,

Non-response is usually one of the "touchy" areas that sparks a lot of
debate in the stat community.  [Mostly because any assumption made is a
complete guess -- and often turns out to be wrong ... but folks do it
anyway.  <shrug>]

{Anecdote: One of the US Census subsurveys used to deal with fruit-tree
growers.  One of the questions was along the lines of "How many trees do
you have?"  They used to allocate the non-response to the same
distribution as the respondents.  Until one day when a researcher
noticed the responses they received were either very large numbers (from
large agricultural concerns) or very small numbers (from mom & pop
growers).  Additional research uncovered the fact that the "middle
market" was *extremely* competitive - which made those respondents more
'close-mouthed' about the number of trees they had than other folks.  So
in this case, making an assumption that the non-response was the same as
the respondents -- was a bad idea.}

Some of the better stat programs, (SAS for one I believe), used to give
you the option to include or exclude non-respondents in the results.

I think we're mostly looking at three options: (a) ignore it, (b)
account for it as "refused/unknown", or (c) allocate it using some kind
of assumption. [Which isn't always a bad thing to do ....]

Being a tad conservative, I think I usually opt for "b" and would report
this more like "56% male, 12% female [46/381], 32% refused". ... or "For
those answering the question; 82% male, 18% female [46/259]".


<confused (&& ||) personal opinion>
Is the surprise or interest or question - "Why the high (32%)
non-response?"  [Personally, I would guess it's "none of your bloody
business" backlash arising from constant privacy invasions through spam
and sug-ing. {Selling Under the Guise of research - e.g. "Can I ask you
a few questions for our survey? Great! ... Can you tell me why you
haven't yet used our amazing coupon for 10% off our carpet cleaning
services available Mondays and Wednesdays in your area?...."}]  Hmmm ...
while I do recall conducting some research indicating that females were
more sensitive about personal privacy (which made them a better target
market for privacy telecomm products like caller id) ... I don't recall
investigating gender specific inclination differences towards
self-reporting their gender ... if that made any bleeding sense at all.

Or ... Is it surprise/i/q because the reported male percentage is so
low?  I think it would be way cool and show some unexpected industry
maturity and perhaps changes in society if our female membership was
higher than traditional 'male-dominant' techie areas would predict! I
think that would be a big plus for the community and elsewhere! 

[Uh Oh!  Does anyone see a need to cross this against a visual arts vs
coding question to see if it holds across the board?  I think THAT would
make a heck of a statement -- 'cuz I think a few of my feathers would be
ruffled the wrong way by any potential interpretation that the gals were
the artists and the guys were the coders ... see whut I mean, Vern?]
</oddly rambling confused semi-rant>


RonL.
(Half remembering some old jape about statisticians being sent out onto
the battlefield to bayonnet the wounded so they could cleanly compute
"survival rates" without dealing with non-response...)


-----Original Message-----
From: spinhead [mailto:evolt at spinhead.com]

Well, that's 44% female; that's not bad, really.

Oh; some of the abstainers could be MALE? Hmmm . . . what a concept.

Statisticians and psychologists: can we assume the same ratio of
abstainers
as there are confessors? Do we do our ratio w/o the abstainers? What's
the
significant knowledge we can get from this?

spinhead


----- Original Message -----
From: "Daniel J. Cody" <djc at members.evolt.org>

> one intersting thing i haven't been able to figure on was 213 of the
> respondents identified themselves as male, 46 as female with 122
> abstanining from the question. not that we need to scramble to find
what
> we can do to get more women invovled imo.