The odds favor Obama gaining the nomination tomorrow
Now some of the bigger pollsters have weighed in and the evidence is mixed for the predictability of tomorrow’s vote. The earlier, in-state polls had indicated the momentum had shifted Clinton’s way. But the most accurate of the big pollsters to weigh in this morning, (per the ratings done by SUSA after Super Tuesday) - Quinnipiac University - indicated the erosion of Clinton support in Ohio was continuing. Consider:
Feb 14: 55% for Clinton
Feb 25: 51% (down 4 pts in 11 days)
Mar 3: 49% (down 2 pts in 7 days)
For Obama, the upward trend is evident
Feb 14: 34% for Obama
Feb 25: 40% (up 6 pts in 11 days)
Mar 3: 45% (up 5 pts in 7 days)
At each Quinnipiac poll, the shift towards Obama has been occurring at roughly 2 to 1 in his favor. At that pace of change, since the margin was 49-45 Clinton through yesterday, with 6% undecided, the trend suggests Clinton will get 48% to 51% of the vote and Obama will get 46% to 49% of the vote. Averaging those ranges, it suggests Clinton will retain Ohio by 2%, though even Quinnipiac was missing by an average of 4.83% through Super Tuesday. If that margin’s consistent, it still says a Clinton win, by about 1/4 of 1%!
(Note: if you’d like to spare any further number crunching, check the last paragraph in this post and see what I’m predicting for both states.)
Continuing with Ohio polls released today, Suffolk had it at 52-40 Clinton, with 8% out. With voters breaking 2 to 1 for Obama, this would suggest an end result of 55-45. Per the SUSA ratings, Suffolk’s margin of error was 7.43%, and since the error margins in Suffolk’s cases undercounted Obama, this suggests Clinton will win by just under 2%.
PPP came in with 51-42 Clinton today, but with its margin of 8.11 pts (also undercounting Obama), it also suggests Clinton wins, by less than 1%. However, note that both Suffolk and PPP show Clinton reversing the trend, while QU doesn’t.
There’s one more, Rasmussen. It also shows a trend reversal favoring Clinton and gives it to her by 50-44, with 2 to 1 breaking her way now. If that trend’s right, the last 6% should project out to 54-46. Yet their margin of error has also consistently undercounted Obama, by 7.61 pts. That would yield a Clinton victory, but again, by less than 1%.
This inexact science yields nothing for certain, but with the numbers available, there seems to be - with the MoE factored in - a consistent read that Clinton will win by no more than 2%. Without the MoE factored in, another crude and mathematically inaccurate strategy of averaging all these poll results (wrong due to different sample sizes and different timespans covered) would put her win around 8%. As I think the former is a more accurate measure, I’ll go with Clinton by 2% in Ohio.
(Update: actually, counting Zogby, whose error margin overcounts Obama’s vote, it would suggest a 4% Clinton win. SUSA also came out with a 54-44 Clinton lead, also showing a trend reversal and with its 4.5 pt error margin, this indicates a 5.5 pt Clinton win. So I’ll adjust my best guess to say Clinton will win Ohio by less than 6%.)
And what about Texas? Back on February 10, I rated a number of demographic factors that displayed the most predictive usefulness. It said both of these states should be Clinton wins, while noting that Obama had recorded a few major upsets above the predictive outcome, picking up CT, ME, WI, that should have gone Clinton. MN was a minor upset for him, too.
That suggested she should be favored slightly to win Ohio, which is consistent with today’s polls. It also indicated she was slightly favored in TX, while noting that the close vote in heavily Latino NM meant an Obama upset was more likely in TX than OH. So what do today’s polls suggest?
Public Strategies had it 46-46 with 8% out. As SUSA hasn’t rated them, its past accuracy is unknown.
Zogby has it 47-44 Obama with 7% out. This projects out to 51-47 Obama with slightly more undecideds breaking his way. With Zogby’s error margin (per SUSA, OVERcounting Obama), this could mean a Clinton win by under 4%.
As for the more reliable polls:
PPP has it 50-44 Clinton, but with its margin of 8.11 pts, this should read a 2% Obama win.
Rasmussen has it at 48-47 Obama. It shows a trend reversal favoring Clinton, with all of the 3% shift going her way (while his percentage is unmoved.) With 5% undecided, if we award ALL those 5 to Clinton, it would project out to a 52-48 Clinton win. But with Rasmussen’s error margin (per SUSA) at 7.61%, this reads as a 4.5% win for Obama.
And SUSA came out with a 49-48 Obama win in today’s poll. It also noted that Clinton had, at least, stopped Obama’s momentum, and may have caused a slight reversal. In fact, compared to its poll 6 days earlier, Obama’s 49% has remained consistent while Clinton picked up 3 pts. So with just 2% undecided, if we gave ALL of it to Clinton, that projects to a 50-49 Clinton win.
But again, SUSA’s error average has undercounted Obama’s margins by 4.5%, so add 2.25% to Obama and subtract 2.25% from Clinton and it works out to about a 3.5% win for Obama.
All three then, project out to 2%, 3.5% and 4.5% Obama wins. As SUSA has been the most accurate of the bunch, my best guess says an Obama win by less than 4%.
Conclusion:
Ohio goes for Clinton by under 6%.
Texas goes for Obama by under 4%.
A few extra points:
1) The key finding in SUSA’s Texas poll was “Clinton’s inroads in the final week come among voters age 35 to 49, where she is even today, after trailing by 18 points last week.” This is consistent with my report yesterday, that the Clinton campaign, instead of focusing on its core supporters (where Obama’s inroads were complete) was targeting Obama’s core - younger voters. Though they had clear success without the media catching on, it still looks, at this moment, like the effort fell short, though it did prevent the blowout victory Obama was trending toward.
2) It appears my conclusions vary little from what Team Clinton’s inside polling shows. Their BEST scenario says a squeaker win in Texas and their Ohio range conforms to mine.
3) Pretty solid ‘True’ ad for Clinton might give them a Texas squeaker? David Corn has a decent rebuttal, but most voters won’t see it. While it could tighten it further, Texas’ early voting, before this last-minute ad - should still give Obama the win.
4) Note that this call by Bill Richardson indicates he’ll support Obama’s claim to the nomination, because Clinton willl not gain more than 10 delegates on Obama’s lead by tomorrow. Typical Richardson performance, though, to not endorse beforehand where his words would have impact on something besides superdelegates.
5) And Clinton will ignore Richardson’s call. If she gets two wins - even if Obama gets a tie in delegates tomorrow - she takes the race to May or June. My projections still say the delegate difference will be 50-50 the rest of the way, so Obama’s lead will hold. Leaving the FL/Mi conundrum and superdelegates to decide this thing.
Unless Obama wins TX tomorrow. Which he should.
