--- In [email protected], Sal Sunshine <[EMAIL PROTECTED]> 
wrote:
>
> On Feb 29, 2008, at 6:56 PM, bob_brigante wrote:
> 
> > People do try to give socially desirable answers when polled 
(they do
> > not see it as lying). Most middle-class whites don't want to give 
the
> > appearance of being wary of racial minorities or of women as 
leaders
> > (even when their names are not going to be published),
> 
> What nonsense.  Anonymous polls are just that.  While a few might  
> lie, most would not, there wouldn't be any reason to.

Oh, sure there would. Here's this pollster they're talking
to, who presumably has his/her finger on the pulse of
the nation, and they want the pollster to associate them
in his/her mind with the "right" side of things. In the
voting booth, in contrast, *nobody* knows who they've
voted for.

It's a known effect, actually, even has a name, the
Bradley effect. There have been several elections
where one candidate was white and the other black, in
which pre-election polls and even exit polls showed the
black candidate winning, but when the votes were counted,
the white candidate won.

http://en.wikipedia.org/wiki/Bradley_effect

Some pollsters and analysts dispute it, but more in
terms of its being big enough to cause an upset
than whether it actually exists.

There was speculation that this is what happened in
New Hampshire, when Obama was widely expected to win
but Hillary beat him solidly. When they looked more
closely at the results, it turned out that the polls had
given him very close to the percentage of votes that he
actually won.

The difference was that the vast majority of voters who
who were undecided at the time the polls were taken broke
for Hillary. So the polls weren't incorrect, they just
hadn't been able to register the size of the vote for her
because people made up their minds at the last minute.


Reply via email to