November 1, 2010

A Truly Bizarre Tale of Two Models

Hatched by Dafydd

A Democratic polling firm, Public Policy Polling, has abruptly shifted its turnout model in their poll (taken largely over the week-end, which typically favors Democrats); what's a bit surprising is the direction and magnitude PPP has shifted in a single week.

In PPP's earlier poll of 10/21 - 10/23, their sample of "likely voters" included 34% Republicans and 47% Democrats, giving Democrats a whopping 13-point advantage; that is, if you'll recall our earlier post -- in which columnist Sean Trende at Real Clear Politics educated us about turnout in the 2006 and 2008 elections and polling turnout models for this year's contest -- PPP's turnout model last week contemplated a Democratic advantage over Republicans even larger than the 12 points they enjoyed in the 2008 presidential election... a Democratic tidal wave!

Consequently, they confidently announced that Sen. Barbara Boxer (D-CA, 100%) was leading Republican challenger and former Hewlett-Packard CEO Carly Fiorina by a strong 9 points; while retread gubernatorial candidate Jerry "Governor Moonbeam" Brown was crushing Republican challenger and former eBay CEO Meg Whitman by a resounding 11 points.

But a whole week has passed; that's a lifetime! (Well, to a mayfly.) In today's release, conducted 10/29 - 10/31, the roster of "likely voters" includes 37% Republicans and 44% Democrats, or a 7-point Democratic advantage; in Trende-speak, that's a turnout model for the current poll just slightly higher than the Democratic advantage in the 2006 (not 2008!) election. From a 13-point advantage down to a 7-pointer in just one week.

Not surprisingly, the leads dropped as well: Today, PPP has Boxer over Fiorina by 4 points; not 9, and Brown leads Whitman by 5 points, not 11. Now that's momentum! (And remember, PPP is an openly Democratic polling company which prefers to poll on week-ends.)

Recall also that all signs indicate Republican enthusiasm is much, much higher than it was in 2006, or even 2004 (when Democrats in California also enjoyed a 6-point turnout advantage over Republicans). In fact, it appears even higher than in 1994; which means, in my moderately informed opinion, that we're very likely to see a much lower Democratic advantage in California even than the 6% of 2004 and 2006 -- as low as 2 or 3 points, or maybe even no advantage at all.

If turnout in this state is at that level, then instead of a Democratic lead of 4% for Boxer and 5% for Brown, this very poll recalculated to those numbers would show both Democrat leads in the 1% to 2% range -- noise, in other words.

And now, the rest of the story.

Another poll was released today, this one from SurveyUSA. It was taken roughly over the same period of time and was also an update, one week later, of a previous SurveyUSA poll. This makes for a perfect trend-comparison with the Democratic PPP poll.

SurveyUSA has also changed its turnout model -- but in the opposite direction, now predicting a greater Democratic advantage than last week:

  • In last week's SurveyUSA poll (10/21 - 10/25), they used the turnout model of an 8-point Democratic advantage over the GOP; this is more than the 2006 election, but still much closer to that than to the 2008 election.
  • But today's release, taken from 10/26 - 10/31, contemplates a Democratic advantage over Republicans of 10 points; this is, of course, significantly closer to the 2008 election turnout than the 2006.

If we believe this poll is honest, we must conclude that everything the SurveyUSA pollsters have seen, coupled with their years of experience, tells them that over the past week, Democratic enthusiasm and eagerness to show up and vote has surged forward! In the last week, they've become almost giddy in anticipation of a monumental, historic victory over the GOP.

Contrariwise, Republicans are increasingly disspirited, apathetic, and beaten down. ("Pay no attention to those so-called Democratic pollsters at PPP; they're all a bunch of crypto-fascist Republican stooges. Progressives rule!")

This is astonishing, so astonishing that we have, I believe, but two possibilities:

  • Either the SurveyUSA pollsters are all dolts, because Democrats have certainly not become more confident and enthusiastic over the last week;
  • Or else this last SurveyUSA poll before the election is simply dishonest.

I cannot believe the statisticians who work at SurveyUSA are that incompetent; draw your own conclusion.

The astonishing part is not that Democrats would be jubilant and triumphant heading into tomorrow's election. Everyone knows they're precisely the opposite and getting gloomier by the hour, and what is obviously false therefore cannot astonish.

But I certainly do find it astonishing that a respected pollster would be so willing to traduce its own reputation and (formerly) good name, to hurl itself under the bus and make itself a laughingstock, all in service to the "great cause": desperately trying to stave off the utter defeat of its allies in the Democratic Party.

It's a sad day. There was a time, in the not too distant past, when polls actually meant something.

Hatched by Dafydd on this day, November 1, 2010, at the time of 5:13 PM

Trackback Pings

TrackBack URL for this hissing: http://biglizards.net/mt3.36/earendiltrack.cgi/4646

Comments

Post a comment

Thanks for hissing in, . Now you can slither in with a comment, o wise. (sign out)

(If you haven't hissed a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Hang loose; don't shed your skin!)


Remember me unto the end of days?


© 2005-2009 by Dafydd ab Hugh - All Rights Reserved