ALL the polls in PA are correct! [w/poll!]
Fri Apr 18, 2008 at 09:23:24 AM PDT
Chris "Tweety-bird" Matthews listened to that impeccable political oracle, his gut, and it told him 14%, the margin by which hillary wins in PA next week. Survey USA, prior to Tweety's gut [which is distrusted by some, hated by others] also predicted a 14% margin. Other pollsters predict a tighter race, from 6% [Quinniapac], 3% [Rasmussen], to +3% [PPP].
Kossacks parse these polls, dismissing some heralding others, in an effort to raise their hopes [PPP] or lower their expectations [SUSA], proving, like the old Billy Joel song, that so many of us vaccilate from sadness to euphoria, spending very little time in between the 2 extremes.
As someone who has studied polling and statistics, AND who has conducted professional polls on a massive scale, I will say right now: it is quite possible that all of the PA polls [minus Tweety] are correct about the 4/22 primary, odd as that may seem.
First, a word on these 'snapshot' or periodic [interval] polls. Campaigns generally employ expensive micro-market daily polls conducted by in-house pollsters. What they are measuring is quite different than what Gallup, SUSA and Rasmussen are measuring.
These interval polls are statistical polaroids of a small piece of an electorate. If PA voters were at a huge party, Obama voters wearing blue, hillary voters in red, and you needed to take a photo of them for the local newspaper, you'd either pan WAAAY back and capture no detail [just the number of how many there might be], or you'd select a subgroup of red and blue shirts, mix them in an assumed ratio, and shoot. The larger the subgroup the more it reflects the actual whole, but the less detail you can discern in the photo. The expense comes in calculating who actually will vote--who belongs in the photo--and how you determine that, by speaking directly to the crowd or through a proxy [robocall], has a large effect on the results.
Interval polls say little about how the universe of participants is changing over time; pollsters and pundits make guesses about why a candidate rises or falls based on events such as debates, controversies, etc., or on how the public's opinion of a candidate is shaped as voters find out more about them: Obama, the greater unknown, is more susceptible to the ups and downs of this since voters believe they 'know' hillary much better.
That SUSA's results are so different than PPP is only surprising if you don't look at the sampling data. SUSA is polling a universe of Democratic voters modeled on 2000 and 2004 data. To the extent that the Obama/hillary contest is similar to those primaries, their predictions could be correct. PPP is polling a universe of voters that may be possible given completely different circumstances--a snapshot where Obama supporters are fighting to get into the picture.
If SUSA is correct, and they might be, since PA is a more demographically static state than most others [people don't move very often, age cohorts are stable over time], then hillary should win by at least 10%. On the other hand, if Obama's possible universe of voters [AA, youth, males, college-educated] is highly motivated and comes out in better-than historical margins, then he should end up within low single-digits of hillary on election day.
If the NCAA finals were on TV that night Obama would probably do poorly. Conversely, if Jerry McGuire and Terms of Endearment were screening on election day in PA, I'd add 2 pts to Obama's totals.
The micro-market, daily tracking polls that the campaigns employ [the much-vaunted 'internals'] are 25MP digital cameras compared to the grainy, but inexpensive, polaroids used by the public polling firms. We rarely are permitted to glimpse them [and usually for dis-information purposes] because they reveal too much of the strategy a campaign employs to court a particular voting bloc. I will simply say that while we Kossacks are in the dark about what will happen next tuesday, the 2 campaigns now know, within 3-4 points, who exactly will be voting next week. HOW they will vote--the true undecideds--is harder to predict.
And that is why these people get the big money while you and I will vacillate between sadness and euphoria from now thru election day.
Permalink | 45 comments