Re: [MLUG - DISCUSSION] [POLITICS] Was the 2004 Election Stolen?

["Jonathan King" <EMAIL:PROTECTED>]

On 6/2/06, Vern Green <EMAIL:PROTECTED> wrote:
> >
>
> I guess the point I was trying to make on a national level could be
> made better with this example.
>
> Since the majority of the people in California live in three main
> cities in California, and since most of the media outlets polling
> would be located in those cities. One would expect those numbers to be
> pretty accurate, especially since the rural voter in California cannot
> even begin to sway the election.
>
> However, on the national level, there are plenty of voters in smaller
> markets and rural areas that can sway an election.

But that's not how exit polling works.  You don't poll cities. You
poll PRECINCTS.  Some of the pollsters are right out there in the
boonies, others are in downtown LA, others still in suburbs of
Sacramento...they're all over.  For a national effort, you have data
from literally hundreds of precincts that you've chosen either
completely randomly or, more likely using a stratified sample that is
picked to be random, but respecting (for example) the mix of
urban/suburban/exurban/rural areas you see nationwide, plus a wide and
even geographical distribution of precincts across the country.

> So if you are
> polling in urban areas, then it is expected that Kerry would be ahead,
> but as the rural vote starts coming in, those exit polls will become
> less reliable

Again, that's not how exit polls work.  You're polling in the morning,
at noon, in the afternoon,  in the evening...attempting to match the
traffic at the polls during the day.  And you're doing it in cities,
in the country, in the suburbs...

Now one thing you're saying is true: if you only go by the morning
polls AND MAKE NO ADJUSTMENTS FOR THAT FACT, then you can end up with
weird biases.  But that's not what happened in 2004, either, as far as
I can tell.

> unless of course you are polling an accurate percentage
> of rural voters as well and I am not convinced this is happening on
> the national level.

But it is.  Because you should remember, they're not just polling for
the presidential race, but for other races as well (esp. state-wide
ones).  Actually, what they problably do is poll in every
congressional district, and select maybe 20 precincts from each of
those, trying to match the overall mix of urban/suburban/etc. It's not
going to be absolutely perfect, but it's enough to ace most of the
contests out there.

> If you look at the map, the urban areas really went for Kerry, but in
> a state like Missouri, and even Ohio, there are plenty of rural voters
> to overpower the larger cities. So if I were only polling in say
> St.Louis and extrapolated the winner from results from St. Louis only,
> then the accuracy of the poll is suspect.

Of course.  Which is why they don't do that.  Really, this is not some
brand new rocket-science technique.  Of *course* you don't just poll
Chicago to call Illinois.  Similarly, you don't just sample from rural
Iowa to figure out what the whole state is doing.

> Which brings us to the electoral college and is the precise reason the
> electorial college was put into place in the beginning, I think it is
> a good idea that makes sense even today.

The electoral college is an undemocratic piece of crock that should be
jettisoned like lots of other things we've gotten rid of since the
18th century.  I mean, why does everybody have this love affair with
the electoral college as some noble institution but yet nobody ever
regrets passing the ammendment that allowed the direct election of
senators?  I mean, it's exactly the same kind of thing.

jking

_______________________________________________
discussion mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/discussion