Mueller wrote:
"A "statistical sample and study" means that one's data is
consciously gathered to be a representative of a population, and the
representative sample is used to draw scientifically rigorous conclusions
about the entire population.The study did not do that, and specifically
disavowed that because it was unattainable at the time. A statistical study
does NOT mean that one counts things or uses the word "quantitative"
somewhere in the study. Since I have done statistical studies
I am perhaps a bit more aware of this distinction than you are."
Thank you for clearing up the distinction between quantitative study and a
statistical sample and study. You have to be patient with me because
unlike all the people on this list and everyone else you cite the study to,
I don't have a PhD (although some of the people I work with professionally
do).
I'm still puzzled about one thing though. Granted that your selection of
cases was admittedly unscientific, then, as you point out, it couldn't be
used to, in your words, "to draw scientifically rigorous conclusions about
the entire population."
Now I suppose you could attempt to draw such conclusions, but such
conclusions would, at best, be guestimates, right? You attemtped such a
guestimate - I quoted the passage from your study, but in your detailed
response to me yesterday you snipped out that part. This is what you said:
"If these proportions are projected into the NSI figures about the total
number of trademark disputes it has handled, one could estimate that of the
2 million or so domain names registered, there are probably only about 257
cases (2,070 x 0.124) of infringement. This amounts to only 0.0128 percent
of all domains."
So that's an estimate about the whole universe (by the way, I don't agree
with your "proportions" for reasons I can get into elsewhere and I don't
believe the NSI figures is even close to being the relevant universe for a
bunch of different reasons).
Now you have done many statistical studies and you are aware of the
difference between a guess and a scientifically rigorous conclusion, and
you indicated that we couldn't draw them form your study (more a survey,
really).
I'm concerned that people who aren't familiar with statistical studies
might jump to the wrong conclusion. It seems that it could lead to a
misunderstanding if you took an unscientific database (not a sample),
analyzed according to your own system, and then projected them against a
subset which you indicate to be the whole. So when your study was cited to
Congress, the DNRC said:
"A recently conducted study by Professor Milton Mueller of Syracuse
University finds that hundreds of domain name holders are losing their
domain names despite only 12 percentage of ALL domain name conflicts
involving trademark infringement. The VAST MAJORITY of conflicts are what
Professor Mueller calls "string conflicts," the types of conflicts over
basic words and names in the examples discussed." (Emphasis Added).
The DNRC must have understood you to mean the whole, because they referred
to hundreds of cases when you only had 59 so-called string disputes in your
"database." And they referred to "all domain name" disputes, so they
certainly thought that they could project your number against the whole.
They didn't say something like, the survey found that 12% of the case
surveyed were ....., they said 12% of all domain names.
Someone else didn't understand your subtle distinction between a study with
quantitative analysis and a statistical study. Back in August someone
posted to the ifwp list, and cited your study for support of the wstatment
that "it is now an ESTABLISHED FACT" that "brand holders who have abused
the NSI process" outnumber "brand holders who have been abused by name
squatters." (emphasis added)
Now, as you point out in the survey, your database was unscientifically
composed (and it included non-NSI cases, if I'm not mistaken), and NSI
doesn't release its numbers, so you really couldn't draw rigorous
conclusions about the NSI black box numbers, so the writer was certainly
under a misimpression. Puzzling, really, because the writer introduced
himself as "an unbiased academic scholar."
This very same writer cited your study to WIPO stating:
"[an] empirical study of trademark-domain name disputes projected, based on
STATISTICAL EVIDENCE, that of the 2 million or so domain names NSI
registered, actual infringement cases constitute a very small number-about
257 cases in the open, generic TLDs" Emphasis Added.
Given that the writer gave such a specific number ("about" 257) I think you
should write him a letter explaining to him how your study couldn't support
scientifically rigorous conclusions or "Established facts" or "statistical
evidence" and that you do a lot of statistical studies so you know the
difference.
The poster's address was [EMAIL PROTECTED]
>
__________________________________________________
To receive the digest version instead, send a
blank email to [EMAIL PROTECTED]
To SUBSCRIBE forward this message to:
[EMAIL PROTECTED]
To UNSUBSCRIBE, forward this message to:
[EMAIL PROTECTED]
Problems/suggestions regarding this list? Email [EMAIL PROTECTED]
___END____________________________________________
