From: James Mahaffy <mahaffy@mtcnet.net>
> I commend the effort that Roy has made, but I am afraid the poll is
> flawed because Roy did not make or modify the questions enough to
> address his own research questions or modify it for a different
> audience. I know there were questions the reflected the original
> pollsters, who didn't understand some of the Christian positions and as
> I indicated earlier should have been reworded. Unfortunately I don't
> think the poll sampled a percentage of any group or even randomly
> sampled any groups. That being the case, one can not validly generalize
> beyond the folks who took it.
I appreciate your comments and agree to a large extent. Let me address the
poll sampling. In the original poll the pollsters selected three
Universities (in California, Texas and the East Coast) [I'm going by memory
here] to conduct the poll. They chose California because Evolutionism and
Creationism were supposedly taught equally in the High Schools. They chose
Texas to try to sample the Bible Belt. Then then had all the students in
several select classes take the poll.
Though they did obtain some 1300 samples, I don't see much randomness here.
Not to mention that the majority of those samples were College aged --
hardly a group on which mature thinking rests and surely not representative
of the entire population.
The sampling on the poll I'm conducting is limited though it is continually
(daily) growing. And I have chosen whom I would sample -- primairly those
who are the most deeply interested in the subject. Just how that reflects
the general population I cannot say, however it should reflect those who are
interested enough to be on email lists such as this one. The sampled have
come primarily from the ASAnet, CRSnet and the SDAnet (which explains why
the SDAs are so out of proportion compared with the actual population). A
smaller percentage has come from a small assortment of 'Christian'
newsgroups. I want to sample Talk.Origins, but I have not personally be
able to announce the poll there because I am unable at this time to access
that group. (If someone would be willing to post an announcement to T.O {or
any other group} please do.)
I don't know if (in the preliminary analysis thus far done) I have attempted
to draw relationships outside of the polled group. But the analysis should
at least reflect the majority of the ASA and CRS and other Christain groups
polled. The preliminary analysis is based on only the first 200 samples
acquired thus far (We now have about 280). I hope that we will acquire over
1000 samples. Then I'll consider a final analysis.
The Questionnaire has been on-line for only 2 months. I hope to be able to
announce the pole to as many on-line groups as I can find and thus increase
the sampling of the on-line population. It takes time.
I have been receiving quite a few comments (such as this by James) about the
questions and the poll. From these and from observing the responses to the
questions, I have been learning how to better prepare a future poll on the
topic. I really cannot make any substantive modifications to the current
one, because that would invalidate all the previous responses and I'd have
to ask everyone to retake the poll. I think the better course is to suffer
through the problems of this one, acknowldge the shortcomings and only draw
conclusions where the evidence appears the most reliable.
Unless you know you have a representative
> or random sample of a population generalities simply can not be validly
> made beyond the folks who responded to the poll.
True. I suppose it depends upon what one's definition of population is. If
we are refering to the general population of the USA, then this will fall
short of the mark. But if we restrict our population to those for whom the
conflict between Creationism and Evolutionism is of concern (those whom most
of the rest of the population hold as authorities), I believe we can apply a
good part of the analysis to them.
> Yes, the results in this sample may be indicative of things that should
> be checked out but I really think few conclusions can be drawn to the
> larger populations of scientists who have different positions on
> origins.
Perhaps. This poll is limited to those with Web access (which I believe
most scientists have access to). It is limited by the lack of an easy means
to announce to all (or most) on-line scientists that the poll exists.
>
> If more work had been done on modifying questions and thinking about
> sampling methodology, the poll would have results that could be very
> interesting.
I agree. I hope to get better at this. :)
Allen Roy
This archive was generated by hypermail 2b29 : Wed Jan 26 2000 - 13:49:24 EST