Friday, September 28, 2012

THE PRAIRIE EDITOR: Counting Chickens

A lot of Republicans have become alarmed in recent days, and a lot
of Democrats elated, by media and poll reports showing Barack
Obama with a growing lead in the presidential race. These reports,
and their conclusions are based almost entirely on poll results, most
of which are plainly (and transparently) overweighted with Democratic
voters (weighting based on 2008 turnout). There is no indication
whatsoever, even by the most optimistic partisan analysts that the turnout
in 2012 will resemble 2008. If anything, it is much more likely to resemble
2010 when the voter intensity was on the Republican side.

There is a double edge to the consequences of these faulty polls. The
intended consequence is to demoralize Republicans and conservatives,
and to stampede undecided and independent voters to the liberal side.
The unintended consequence, however, might be to make Democratic
voters overconfident and to diminish their energy in the remaining days
of the campaign.

As I have pointed out many times. pollsters can “play around” with the
numbers, either out of bias or ignorance, rather freely when the election
is many months or weeks away. Sheer self-interest and survival instincts
reduce this tendency, however, as the election itself approaches. No pollster
wants to be humiliated by being on the record with a ridiculous poll just
before the election.

I am speaking here of media polls, that is, polls that are conducted primarily
to be very public news events. There is another kind of polling going on
simultaneously by campaigns themselves, usually referred to as “internals,”
which are rarely reported, but which serve as guides for candidates and
their campaigns about how they are doing. These are much more expensive
polls, and are weighted very realistically. No campaign is going to pay a lot
of money for an internal poll that gives them a false picture.

There are many more polls, particularly state and national ones, in 2012
than in previous cycles. The all-important sample number varies widely.
The weighting (which is simply adjusting the raw results) of a poll sample
by party varies even more erratically. Many pollsters and their polls are paid
for a by a political party. It is clear that reporting poll numbers has become
part of the “warfare’ of a political campaign. In my opinion, few polls
should be taken very seriously because few polls are successfully trying to
avoid the bias that comes from bad weighting,  inappropriate technological
inquiry procedures (such as using only land-line telephones and not cell
phones), or from the statistical consequences of repeated sampling to get a
response.

One of the few national polls which seems to be trying most fastidiously to
reach an accurate result is Rasmussen. Their methodology seems to be
the most energetic to avoid a distorted result. While Gallup and other national
polls are showing a 3-6 point margin for Obama currently, Rasmussen is
showing it be either an exact tie, or depending on the day, a one-point
margin for either Romney or Obama. These are simultaneous polls, so
someone is wrong.

Whether or not Mr Romney has fully “sold” his point of view to voters
can be debated, as can the impact of his “47%” video remarks, but there
has not yet been any real evidence presented that his campaign is “falling
behind.” It might be true, on election day, that he will fall short, and that
Mr Obama would be re-elected. Similarly, there is no real evidence yet that
the president’s campaign is certain to fail. The presidential debates are ahead,
and I suspect that they will be more significant than usual in this campaign
cycle. Voters already know Mr Obama, but many do not know Mr. Romney,
especially standing next to and confronting his opponent,

Finally, the current poll distortion, if it is that, offers a greater danger to the
Democrats than to the Republicans. With less than six weeks to go, a mood
of overconfidence, provoked by currently reported poll numbers, could easily
be transformed into utter panic for Democrats if, as election day approaches,
the polls are reversed favoring the Republicans. As any experienced political
observer will tell you, momentum is a huge force just before and on election
day.

Every pollster, good or bad, will say that a poll is only “a snapshot in time.”
But there are snapshots, and there are snapshots! That is why good cameras
cost more than cheap ones.

No one should think this campaign is over, nor that it is in a final trend, nor
that the information they are receiving via polls is accurate. Much more lies
ahead, including most importantly, what we will see when the two presidential
candidates are in front of us together.

-----------------------------------------------------------------------------------------------
Copyright (c) 2012 by Barry Casselman.  All rights reserved.

No comments:

Post a Comment