Get yer polls, bets, numbers & pretty graphs! Elections 2008

Any predictions for today?

My prediction is that Obama wins NC and Hillary wins IN (so far so obvious) -- and that both will win with about the same margin. I dont know how high the margin will be (5-10%?), but it will be roughly the same in both states.

Better than that would be great, of course, especially with the prospect of the unprecedented battering Obama will face in West-Virginia and Kentucky (nothing to do about that); a cushion would be good.

When it comes to those WV and KY races all I can hope is that the press will deal with them roughly like they did with the race in Mississippi - so obvious an outcome that it's not really treated like news, let alone some kind of game changer.

I will stick my neck out and predict Clinton wins by a wider margin in Indiana than Obama does in NC. I am basing it upon Obama doing worse than expected or projected lately. I doubt Clinton wins NC, but it could be closer than expected.

Here's the last North Carolina poll from each pollster that still conducted at least one in May:

NORTH CAROLINA



Pretty broad consensus really. There's six polls - if for both Clinton and Obama you disregard the highest and the lowest poll result and look at the four in between, you have 40-43% for Hillary and 49-51% for Obama.

There's a fair number of undecideds though: except for SUSA and ARG, these polls have 7-11% undecideds, which is more than enough to seriously change proportions. So what's important is what way the undecideds break.

Hillary has been doing well with last-day deciders, especially lately, so probably the 5-10% Obama lead that the consensus here seems to point to will narrow somewhat. But then Hillary has been winning primaries lately, and an alternative theory is that undecideds just break toward the winner - I remember that in the beginning of the primaries, back in Iowa and South-Carolina, it was Obama who picked up a "winner's bonus" and outdid the polls.

It might also be worth looking at what the trend has been. Zogby, PPP, ARG and Rasmussen all had Clinton closing the gap compared to late April; IA and SUSA had it stable. On the other hand Zogby had Obama extending his lead again right in the last couple of dates, but then its poll is a bit of an outlier compared to the others. All in all I think the trend also suggests that the outcome will be a bit tighter than the polls might indicate.

I could see Obama winning by 8% in NC. That is down a lot from a few weeks ago. Clinton's folks had a good strategy: send Bill out into the little towns where the major media aren't and where the locals have never seen any national politician before. Keep him from putting his foot in his mouth.
In IN, I see Clinton by 8%. 200,000 newly registered voters, the open primary (independents and Republicans to play), and the proximity of IN to IL could help Obama enough to make it close.
When the dust settles, there will not be a clear winner. So on it will go.
See yall later this evening for the results.

I say NC goes Obama by 4% which should be considered a moral victory for Clinton and Indiana goes 8% for Clinton which will also be a Clinton victory, both morally and actually.

My guess is that NC will be closer to 14% then 4%.

Cycloptichorn

Last-minute overview of what the last polls out said about...

INDIANA



Again, six different pollsters that released polls still in May - same ones as in NC except for Suffolk taking the place of Rasmussen.

Using the same technique of disregarding the highest and the lowest poll number for each of the two candidates and going on the four in the middle, we again have a consensus on Obama's position: he's at 43-45%. Even disregarding the top and bottom polls, however, Hillary's numbers vary more: 48-53%. Relevant detail: the lower the share of undecideds in the poll, the higher Hillary polls.

That's cause for a warning. The same happened in Pennsylvania, and if I recall correctly Ohio too. Obama's numbers stable in the low- to mid-fourties, Hillary's numbers varying widely, and the more pollsters pushed undecided voters for an answer (like SUSA does), the higher Clinton's numbers. In both of those cases Hillary ended up maximising her winning margin on election day.

Obvious explanation would be that either, undecided voters (in the states at hand) tend to break for Hillary when finally forced to make a choice; or that a number of people say they're undecided when they've really already kind of made up their mind for Clinton. Maybe it's a feeling of social pressure that stops them from saying so. Perhaps a manifestation of the Wilder/Bradley effect, which describes a share of voters not telling pollsters that they are going to vote against the black candidate, which results in the black candidate getting roughly what he's been polling for but the white one outdoing the polls on election day.

(Impossible to identify to which extent different motivators are at hand here, really - whether it's undecideds breaking to the 'safe', long-known candidate, or a undecideds breaking along with the trend to the winner, or whether it's pseudo-undecideds who will vote against the black candidate.)

Looking at what the late trends have been, it's a bit of a mixed bag. IA and PPP have Obama narrowing the gap, and Zogby tentatively agrees. But SUSA, which has had a formidable record this year, saw Hillary extending her lead a little, and ARG tentatively agrees. All in all Hillary's lead is set at 4-8% by four of the six pollsters; outliers are Zogby (+2 for Obama) and SUSA (+12 for Clinton). We've seen those two pollsters at opposite ends of the predictions before (California, Missouri and Ohio come to mind). So far SUSA's gotten it right more often.

So, according to the pollsters it will be Obama +5-10 in North Carolina, and Clinton +4-8 in Indiana. But there's a number of things here that plead for expecting Clinton to outdo the polling. Hence why I'm betting their respective leads will be pretty much the same, but why Okie might well be right predicting Hillary will actually bag the bigger lead. Of course NC has more delegates, so in terms of delegates Obama will probably draw the long end anyway, which means both candidates will have plenty excuses to proclaim victory.

from Kos

pdate: Ahh finally more exit poll numbers:

North Carolina: Economy the top issue:

Obama: 52
Clinton: 44

Indiana: Economy the top issue:

Clinton: 54
Obama: 45

Update II: More exits:

Indiana: Vote by ideology for Obama:

Liberal: 54
Moderate: 46
Conservative: 38

North Carolina: Vote by ideology for Obama:

Liberal: 62
Moderate: 55
Conservative: 42

Hillary's slide to Republican tactics and talking points are paying dividends. With conservatives.

* Permalink ::
* Discuss (278 comments)

Exit poll data
by kos
Tue May 06, 2008 at 02:49:19 PM PDT

Per CNN:

Indiana

African Americans: 14% of the vote

Obama 92
Clinton 8


North Carolina

African Americans:

Obama 91
Clinton 6

I forget where I saw it but the view that Hillary does better (or Obama worse) than she polls is not backed up by the overall statistics.

The morons who voted for Hillary because of her gas tax scheme deserve her although IMNSHO she has zero chance of winning the GE if she somehow steals the nomination. Even John McCain can win when African-Americans stay home in droves.

Yesss... this was the day before the Pennsylvania primaries:



And a couple hours later:



Compare now for Indiana:


The similarities in numbers is positively eerie. Sure makes it seem like the Indiana result will resemble the Pennsylvania one very closely.

After being burnt so many times, I've finally learned my lesson: I'm staying as far away from those early exit poll data as I positively can. Won't get led astray again.

The only difference I see is that Hillary was way up for a long time in PA, and they've been closer in Indiana. What I've thought is a possibility is that people have a preconception, are either convinced (Clinton lead reduced) or waver, and the waver-ers tend to go back to their preconception.

Since early Indiana polls were better for Obama (variable but closer), I think it's possible that some of these waver-ers will go back to their preconception but that it'll be Obama.

But that's a spare thought.

Can't tell.

2 minutes 'til polls close in Indiana. Ack.

Too close to call.

WHEW.

They did that in PA too but still whew.

Indiana Exit Poll:

Obama wins under 65, Hillary in a landslide 71-29 over 65.

I am praying they call NC for Obama the minute the polls close. If the African-American number is correct, woth a large turnout, a landslide is very possible.

Much bigger whew!

Hope it's a big ol' margin...

Yay!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!


With the black vote gping 92-7 for Barack Obama and a delegate victory now assured, to let Hillary steal the nomination would be suicide!

Nets call NC for Obama at the close.

Cycloptichorn

Depends. Lately she certainly does. But is it inherently Clinton doing better than polled for - you know, undecided voters breaking for the safer choice, maybe a bit of Bradley/Wilder effect? Or is it just the winner of the primary outdoing the polls - a winner's bonus, sort of?

RealClearPolitics.com has one of the better sets of tables with all the available polling data by state. And unlike the more cautious Pollster.com, which deals only in trendlines, they do running averages, which means that for each state you can see what the average Clinton or Obama lead was in the very last polls that appeared. See the page for Pennsylvania as example.

Mining those pages for a quick overview, you see Hillary repeatedly outdoing the polls lately:

[u]Pennsylvania[/u] Final Results, Clinton +9.2 Polls average, Clinton +6.1 [u]Ohio[/u] Final Results, Clinton +10.1 Polls average, Clinton +7.1 [u]Texas[/u] Final Results, Clinton +3.5 Polls average, Clinton +1.7 [u]Rhode Island[/u] Final Results, Clinton +18.0 Polls average, Clinton +9.7

Clinton outdoes the polls every time! But these are also all states she won, so we're still no wiser whether it's something about her vs Barack, or more like a generic winner's bonus.

Go back further in time:

[u]Wisconsin[/u] Final Results, Obama +17.4 Polls average, Obama +4.3 [u]Virginia[/u] Final Results, Obama +28.2 Polls average, Obama +17.7 [u]Maryland[/u] Final Results, Obama +23.5 Polls average, Obama +22.3

Now it's Obama outdoing the polls every time! A winner's bonus, then?

Back to Super Tuesday:

[u]California[/u] Final Results, Clinton +9.6 Polls average, Obama +1.2 [u]New York[/u] Final Results, Clinton +17.5 Polls average, Clinton +17.2 [u]Illinois[/u] Final Results, Obama +31.5 Polls average, Obama +33.0 [u]New Jersey[/u] Final Results, Clinton +9.8 Polls average, Clinton +7.7 [u]Massachusetts[/u] Final Results, Clinton +15.4 Polls average, Clinton +7.0 [u]Georgia[/u] Final Results, Obama +35.3 Polls average, Obama +18.0 [u]Missouri[/u] Final Results, Obama +1.2 Polls average, Clinton +5.7 [u]Tennessee[/u] Final Results, Clinton +13.3 Polls average, Clinton +13.0 [u]Arizona[/u] Final Results, Clinton +8.8 Polls average, Clinton +6.0 [u]Alabama[/u] Final Results, Obama +14.1 Polls average, Clinton +1.2 [u]Connecticut[/u] Final Results, Obama +3.1 Polls average, Clinton +4.0

And the pre-Super Tuesday states:

[u]South Carolina[/u] Final Results, Obama +28.9 Polls average, Obama +11.6 [u]Nevada[/u] Final Results, Clinton +5.5 Polls average, Clinton +4.0 [u]New Hampshire[/u] Final Results, Clinton +2.6 Polls average, Obama +8.3 [u]Iowa[/u] Final Results, Obama +7.8 Polls average, Obama +1.6

Time to test the hypotheses.

Obama/Clinton

Margin in the final results was at least 1% more favourable for Clinton than it had been in the last polls in 11 states

Margin in the final results was at least 1% more favourable for Obama than it had been in the last polls in 9 states

No significant difference between result and polling in 2 states.

Winner/loser

Margin in the final results was at least 1% more favourable for the candidate who had been ahead than it had been in the last polls in 14 states

Margin in the final results was at least 1% more favourable for the candidate who had been behind than it had been in the last polls in 6 states

No significant difference between result and polling in 2 states.

Conclusion

You're right, Hillary has not been outperforming the polls significantly more often than Obama. That's just what's happened in the last batch of primaries (or at least the primaries that actually had sufficient polling to say something about it, which excludes MS, WY, DC and VT). That might signify a new trend, but since Hillary won all those states as well, it could equally well just be a question of there being a winner's bonus.

There has been such a winner's bonus in 14 out of the 22 above states: a little less than 2 out of 3 times. Still hardly a foolproof trend. But it bears mentioning that the 8 exceptions were 7 Super Tuesday states and New Hampshire. A lot of unexpected reversals on Super Tuesday (CA, MO, AL, CT). In the states since Super Tuesday the winner's bonus has been a given.

So maybe Clinton's lead in the end results, when all the counting's done tonight, will be bigger than polled for in Indiana, but Obama will get a bigger lead in North Carolina than polled for?

Ah, very cool. Writing that post has helped me escape the nervous-yet-meaningless first hour after the polls close.

Even cooler: the results coming in now (North Carolina called for Obama straight away, Indiana showing a 14-point lead for Hillary with 26% of precincts reporting) seem to conform exactly what I wrote in my last para just now