Re: Poll: 1/3 of UK Teachers Believe ID Should be Taught in Schools

On Nov 14, 9:50 am, Greg Guarino <g...@xxxxxxxxxxxxx> wrote:
On Fri, 14 Nov 2008 04:44:50 -0800 (PST), tgdenn...@xxxxxxxxxxxxx
wrote:



On Nov 12, 12:58 pm, Ilas <nob...@xxxxxxxxxxxxxxxx> wrote:
Greg Guarino <g...@xxxxxxxxxxxxx> wrote innews:0n3mh4p4jp5i32vp89v2v52qu6pguq9kep@xxxxxxx:

On Wed, 12 Nov 2008 10:35:25 -0500, "Steven L."
<sdlit...@xxxxxxxxxxxxx> wrote:

Teachers TV surveyed 10,600 education professions and received 1,210
responses, reports The Guardian.

I forget the proper name for the sort of "poll", in which the sample
pool is composed of whomever cares to respond. But if that's what was
done here, it's useless.

10,600 questionnaires were sent out, 1210 responded (self-selected),

You haven't forgotten - it's a self selecting poll

and 351 of those said ID should be taught. From this "data" Teachers
TV concludes that 29% favor teaching ID. Teachers TV needs some
lessons in how to conduct a poll.

If you think about it, the polls for the just-completed US elections
were of exactly the same kind, since response is voluntary. Those
polls were remarkably accurate in aggregate.

The election wasn't "accurate". What would that mean? An election is
designed to select candidates favored by a majority of those who
choose to vote. It does not need to be generalizable; if 99 people
vote, whoever gets 50 votes wins, even if thousands who chose not to
vote disagree. This sometimes happens, for instance, in lower-level
elections, like for school board members.

I think what you mean by "accurate" is that the polls taken to predict
the outcome of the election agreed well with the actual election. But
you will note that those polls take great pains to select "likely
voters" to sample.

The kind of "poll" we are talking about has an entirely different
purpose than an election; to answer a question about public opinion
without asking every single member of the public. Unlike an election,
we *do* want to be able to generalize from whatever sample we take to
the whole population.

One of the requirements then, is to pick a representative sample. That
can't be done through self-selection. You could argue that a properly
representative sample might result by sheer coincidence, but  the
chances of that are even worse than random; the people who care enough
about an issue to spend the time to respond are unlikely to be
representative of the vast "middle". But even if the poll somehow lit
on the "true" number, how would we know?

It is certainly *possible* that the ID sample was self-selected; as
the Wiki article mentions, strong opinions and knowledge of a subject
tend to produce higher response rates. However, you cannot assume that
based on the evidence given here, just because you don't like the
outcome.

What I said was  "IF that's what was done here, it's useless".  I
don't know what th outcome of a proper poll might be, but I can't
think of a way in which this could have been a proper poll. They
"surveyed" 10600, and got 1210 responses.

It would be interesting to know what other information was gathered
about the respondents, since then you might be able to make a case
about self-selection one way or the other.

It's hard to imagine that any particular care went into selecting even
the initial 10600, if they were to be allowed to self-select
afterwards. For one thing, the 10600 were "eduction professions" (I
imagine that means "professionals"), which probably means teachers of
all subjects, not just science. I suppose it's reasonable to assume
that a History teacher is better educated than the public at large,
but not necessarily about science. Could 29% of teachers of all
subjects (including early grade "general" teachers)  favor some sort
of "teach the controversy? Maybe so. But does this poll shed any
reliable light on that question?


Of course it does. You seem to be confused about the term self
selection and polling in general.

Self selection is only a problem (it distorts the correlation between
the sample and the population being sampled) if the self selection *is
connected to the variable under examination*.

In the US elections, for example, some previous data indicated that
less educated people were less likely to answer polls than more
educated people. Since less educated people were more likely to vote
for McCain, it was suggested that the polls might under-represent
McCain voters relative to the population. When pollsters publish or
supply their data, they might correct for this, or they might give two
results with and without the correction, as they see fit---that's what
they get paid for, which is inferring the characteristics of the
population from the sample.

In the ID case, any suggestion of self selection in the sense that it
would affect the outcome is pure conjecture. So you have no reason to
assume that this isn't a 'proper poll'. BTW, 1210 responses is way
more than enough to get an accurate result.

-tg


Greg Guarino

.

Relevant Pages

  • Forward to your MP?
    ... have worked as an election judge. ... poll books that we used to check in voters. ... to all of the voter registration information. ... We set up those machines in a daisy chain fashion, ...
    (uk.misc)
  • Re: What are your plans for election night? (nbc)
    ... 2004 when he was touting a good chance of a Kerry landslide 5 months ... Why the next election won't be close. ... with President Bush and Sen. John Kerry splitting victories ... And poll after poll show the two candidates oscillating within ...
    (rec.music.artists.springsteen)
  • Re: Darwin, Britains Hero, Is Still Controversial In U.S.
    ... and how many had to be told who Darwin was before ... that was why I went to the HARRIS POLL website. ... of the accuracy of the responses of "Yes, ...
    (rec.collecting.coins)
  • Re: Darwin, Britains Hero, Is Still Controversial In U.S.
    ... that was why I went to the HARRIS POLL website. ... from individuals who already were familiar with Darwin and his work. ... designed to steer responses in one direction. ...
    (rec.collecting.coins)
  • Re: For all you idiots that call for a boycott of Israel...
    ... The same type of activity was investigated involving the poll involving Finkelstein's tenure. ... The fucking yids flooded that from very few IPs, ... And I also noted the following responses coming from this IP address 68.127.36.52 ...
    (soc.culture.israel)
Loading