Re: [ga] let's try to see this constructively... and not over rea ct ...

[Kent Crispin <kent@songbird.com>]

On Fri, Jun 15, 2001 at 08:09:51PM -0400, Jonathan Weinberg wrote:
>          I'm going to post one more message in this thread, and then I'll 
> stop, although I have the disquieting feeling that I'm failing to 
> communicate what seems to me to be a simple point.

To me it appears that you are failing to communicate because you
apparently aren't aware of the strong negative bias you exhibit in the 
process. 

[...]

>          [2] I think that people should do whatever they can, within the 
> four corners of the survey, to communicate their views, including using the 
> narrative boxes to the extent they can.  (Alternatively, I know one person 
> who wrote a narrative statement and sent it in *instead* of the survey.)

An option that has always been available, even before the survey was 
announced....

> [3] None of this renders it off-limits to point out that the 
> survey is slanted, and the existence of narrative questions doesn't negate 
> the problem.  If I were to receive a multiple-choice survey in which the 
> first question asked whether the President/Prime Minister's performance was 
> (a) outstanding; or (b) above average, and gave no other choices, than the 
> fact that question 20 allowed me to insert narrative comments would not 
> negate the bias.

If there were such questions, then you might have a point.  But 
Froomkin's contorted analysis is a very far cry from what you describe 
above. 

> >I think you miss Michael's point, Marilyn.  He's not saying that
> >the study is statistically invalid because it allows for narrative
> >responses.  He's saying that it's invalid because it's systematically
> >biased in favor of one set of results.

This survey is statistically invalid no matter how the questions are
worded.  It is BORN statistically invalid, because there is no
meaningful definition of the sample involved.  This is a survey SEEKING
IDEAS; it is not a surrogate vote. 

> >Thus, for example, he mentions
> >Question 9, which asks whether each whois data element is "essential,"
> >"desirable" or "valueless."  Problem is, almost any information has
> >value.  Nobody would say that a registrant's postal address, say, is
> >"valueless" -- but the survey gives no opportunity to say that it is
> >nonetheless unnecessary, or that the privacy costs associated with
> >including it outweigh that value.  Rather, the respondent's only other
> >choices are that the information is either "essential" or
> >"desirable."  That's a slanted question.

The question isn't whether the information is valuable in the abstract;
the question is whether the information is valuable to *you*, the person
filling out the survey, in your use of whois.  The semantic contortions
you and Froomkin go through here are far more indicative of a bias
towards negative criticism than it is of a bias in the survey. 

I am not trying to claim that the survey is perfect -- it *obviously*
isn't perfect.  But perfection is simply not required for the task at 
hand.

-- 
Kent Crispin                               "Be good, and you will be
kent@songbird.com                           lonesome." -- Mark Twain
--
This message was passed to you via the ga@dnso.org list.
Send mail to majordomo@dnso.org to unsubscribe
("unsubscribe ga" in the body of the message).
Archives at http://www.dnso.org/archives.html