<br><br><div class="gmail_quote">On Wed, Jun 23, 2010 at 7:34 AM, BillK <span dir="ltr"><<a href="mailto:pharos@gmail.com">pharos@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On 6/23/10, Stathis Papaioannou wrote:<br>
> Yes, but I wonder how many people really believe the answer they give,<br>
> rather than just saying it because it is socially acceptable.<br>
><br>
><br>
<br>
</div>Well, that suggestion could invalidate every poll ever done. :)<br>
<br>
I believe that Pew works with panels of people that volunteer to<br>
answer surveys for them and they provide Pew with details of their<br>
lifestyle, so that Pew can make a balanced selection for each poll.<br>
The survey is done by phone, so there is no peer pressure to saying<br>
yes, no, maybe, to each question asked. If there was anyone else in<br>
the room, they wouldn't hear the questions. It is also unlikely that<br>
people would volunteer to do surveys in the first place if they<br>
intended to lie. Why bother?<br>
<font color="#888888"><br>
<br>
BillK<br></font></blockquote><div><br></div><div>This particular survey at least seems to have been a random sample, which is weighted after the fact to reflect more representative demographics. see: <a href="http://people-press.org/methodology/">http://people-press.org/methodology/</a></div>
<div><br></div><div>Anyway, while I share your doubt that misrepresentation had a significant effect in this case, but wouldn't say there is no peer pressure over the phone if a respondent suspects the survey taker shares the views of those they frequently interact with (most probably do automatically). There's some evidence that automated telephone polling in which questions are answered by touch-tone to a system that is obviously a machine inspires more honest responses than ones handled by humans (unfortunately this is counteracted by the inability of most automated systems to evaluate the credibility of a respondent, call cell phones, or explain questions, leaving about the same net accuracy).</div>
<div><br></div><div><br></div><div>In other news, apparently 81% of Americans believe ordinary computers will be able to pass a turing test: ("Computers probably/definitely be able to converse like human beings in 2050."). </div>
</div>