Proof Of A Poll’s Propensity To Predict Presidential Elections Precarious, At Best
“Well, it’s just like - they’re people I work with, and our job is being popular and shit.” - Winona
Polls measure popularity. Period.
They don’t predict presidential primaries any more than they predict elections. That’s because they involve people. Real people. People who watch TV. People who read the paper. People with horrible memories of high school or wonderfully improbable dreams of the future. People who can and more-than-likely will, without any notice, without any reason, change their mind.
(I dare anyone out there to defend the accuracy of a poll taken nearly two years prior to a presidential election.)
Still, they keep coming. Poll after poll. These polls are, if nothing else (like accurate), quite popular.
Certainly the popularity of these presidential polls can be partially attributed to an interest in what other people are thinking. But, primarily, the popularity of polling is the product of two particular properties.
- polls are simple/inexpensive to conduct
- polls are, compared to other methods of prediction, easy to understand.
The problem with polls is much more general- and is the same problem faced by all popularity contests: the winner depends entirely on who you ask. Here are some recent examples…
AMERICANS
The latest Newsweek poll was based on interviews with 1003 American Voters- who selected
OHIOANS
The Quinnipiac poll released today was based on interviews with 1,305 Ohio Voters. It shows Clinton (38%) in the lead for the Democratic nomination- followed by Obama (13%) and
OHIO BLOGGERS - LEFT
A recent poll on BSB, based on votes from 197 lefty bloggers, shows strong support for Edwards (31%) followed by Obama (14%), Gore (11%) and Clinton (10%).
OHIO BLOGGERS - RIGHT
The results over at RAB, based on 311 votes from conservative bloggers, have Giuliani (21%) with a slight lead over Gingrich (18%) followed by Ron
See what I mean?
