Archive for September 18th, 2004

Proverbial Ten Foot Polls


The wild
swings in recent polls suggest the race is volatile. The nonpartisan
Pew Research Center and Harris Interactive polls this week indicated
that the race is essentially tied, while a new Gallup survey indicated
that Bush holds a lead of 13 percentage points over Kerry. A New York
Times/ CBS poll of registered voters published today indicated that
Bush leads Kerry, 50 percent to 42 percent.

from today’s Boston Globe

To the Dowbrigade, the widely divergent polls released this week, showing
everything from a 14 point Bush landslide to a virtual dead heat, to
a slight Kerry lead don’t suggest a volatile race.

Rather, they suggest social scientific ineptitude at best, and more
likely deliberate efforts to mislead and influence the electorate.

After all, public opinion polling, although complicated, is supposed
to be scientific.  Professional pollsters will wax lyrical about
the difficulties of their task, selecting a representative sampling,
people deliberately joking with or lying to pollsters, determining who will actually
vote come election day, but these are just mealy-mouthed excuses.

The truth is, given the importance of accurate information in this area and the growing sophistication
of our computer models, modern polling SHOULD be able to predict the outcome
of an election with increasing accuracy. After all, the predictions are
all on record, and in every case culminate in the actual election results,
which can theoretically be used to evaluate the predictions, fine tune
the polling procedure, adjust the methodology used, and produce a series
of ever
more accurate polls.

We should be able to identify and reward those who are making accurate
predictions, and count on their ever-increasing accuracy.  The
fact that the current crop of supposedly professional public opinion
are so all over the map seems to be direct evidence of perfidy and/or
lack of ability, and leave the poor public more in the dark than before,
only now armed with sharp and dangerous statistical weapons.

The only thing all the polls agree on is that the far majority of the
voting public have already made up their minds. Why is it so hard to
find out what they think?

from the Boston Globe