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

[DPF <david@farrar.com>]

On Thu, 14 Jun 2001 21:39:24 -0700, Kent Crispin wrote:
>On Fri, Jun 15, 2001 at 03:50:14AM +0000, david@farrar.com wrote:

>>> Why not use the narrative questions to include other perspectives that you
>>> think aren't included, as informational points?
>> 
>>This can certainly be done by the individuals who think of this but it still 
>>means that the quantative data would have considerably less value than could 
>>have been the case.
>
>In fact, there is very little significance that can be attached to the
>quantitiative data.  Right now people are treating the survey as a vote,
>and busy trying to stuff the ballot by getting lots of people who think
>like they do to fill out the form.  There is essentially no statistical
>validity to that as a sampling method. 

This is true for any self selecting sample.  

>> Both of the above allow each data element to be given an average score which is 
>> useful for compararative purposes.
>
>Without a rigorous sampling methodology, such finesse is useless.  A
>statement like "67% of those surveyed think that whois is essential" 
>would be misleading, at best.

One reports the results noting the sample is self selecting and also
that multiple voting is potentially possible.  However one could well
find a lot of useful information in the results.  It may also
challenge assumptions people have made about the importance of certain
aspects of whois.

>> Question 9 as written is not without value.  One can compare the % who mark 
>> essential for each data element as a rough guide and ignore how the remainder 
>> is split between desirable and valueless.
>> 
>> As with others I commend ICANN/NC for doing such a survey.
>
>While I think your emphasis on the numerical aspects is completely
>misguided, I also agree that the survey is a very positive development. 
>While there is no doubt that it is essentially useless from a quantative
>point of view, it remains true that a great deal of useful qualitative
>information will come from such a survey.

If the question responses had been less leading then the quantitative
data would have been more useful.  It would still have restrictions as
to its usefulness but if you know what those restrictions are you can
still find use for the data.

DPF
--
david@farrar.com
ICQ 29964527
--
This message was passed to you via the ga-full@dnso.org list.
Send mail to majordomo@dnso.org to unsubscribe
("unsubscribe ga-full" in the body of the message).
Archives at http://www.dnso.org/archives.html