Lists Home |
Date Index |
Tony Lavinio wrote:
> Here is the methodology we used:
> 1. We emailed over 75,000 IT professionals, and also had the
> survey posted on our website. We obtained email address from
> tradeshow lists, direct marketing lists, registered customers
> and our entire customer prospect list. We did not as some
> have suggested ask just the XQuery SIG :).
> 2. 550 completed surveys were submitted.
> 3. Of the 550 people who submitted complete surveys, 477
> indicated that they currently use or plan to use XQuery this
> 4. This is a huge sample size... to put things in perspective,
> last year, media news outlets were trying to determine the
> outcome of the US presidential election using smaller sample
Surely you're kidding. A survey is not a sample, and you cannot
treat the results of one as the other. And a serious results analysis would
put the numbers into doubt, rather than play them up.
1) Your initial group was not random
2) Your respondents were self-selected
3) If there were any inducements, such as freebies, then the
respondents would be skewed towards uptakers (I don't know
if any freebies were offered, perhaps Tony could confirm they
4) I take it this is a survey is of Americans, or of English speakers
5) That 7.4% of your respondents have been using XQuery for 2 years
or more speaks for itself. Lets say there are ten million "software
professionals" in the world. Does anyone believe that there were 740,000
XQuery users 2 years ago? Or even 740?
Lets say I have a list of Australian pyromaniacs. Then I ask them
"How many of you have or intend to set fire to a kangaroo
this year?" and I receive several hundred responses, most of which
are "yes". Would I be justified in warning the Canadian police department
that "Most pyromaniacs intend to set fire to kangaroos this year"?
The funniest line in the PDF is "The high XQuery adoption rate it
difficult to dismiss given the quality of the data." ROTFL
I recommend Tony get the material withdrawn, and reissue it with
the press release "Survey shows at least 2/3 of a percent of company's mail
list are people who might find that company's product useful" :-)