Over the period of April 12 - July 1, 662 respondents to the Daily Kos/SEIU/PPP State of the Nation poll were reached at a Texas phone number. Among these respondents, Obama and Romney were tied 47-47. (Please note that this is an aggregation of interviews from the weekly poll, not a stand-alone poll.)
Seriously?? Yes, but it's highly unlikely to hold up until election day, for reasons discussed below, and it's more favorable to Obama than other polls.
The two stand-alone polls from the same time period also show a relatively tight margin among registered voters. Texas is a Tossup at least by New York Times standards, based on this result and the two other polls with Romney +7 and Romney +8, for an average of Romney +5. (New York Times lists CO, NV, and PA as Tossups, with a polling average of Obama +5 or more.)
Of course, this is a poll of registered voters. Typically, the demographics shift towards Republican voters by a few points among actual voters in Texas.
How did this happen? It's not a demographic shift compared to 2008, but rather Romney's lack of popularity among whites. While Republicans have solidified behind him (even those who don't like him) Independents and Moderates have shifted towards Obama.
Similar numbers were seen in early 2008, but it didn't last. By summer 2008, McCain had pulled solidly ahead. So keep that in mind. But it does seem reasonable to think of Texas as Lean Red instead of Solid Red for now.
Details below, including playing with numbers to see what it would take for Obama to win in November. What is interesting is that support from whites doesn't need to be nearly as good as we see in this poll for Obama to outright win Texas, if the Texas Latino community voted at the same rate as Latino communities in Florida or California.
Is this polling aggregation way out of bounds? I looked at demographics, other polls, and neighboring states to see.
The demographics were about on target. There were slightly too many Democrats compared to 2008 exit polls, which makes sense, as this is registered voters. There were slightly too few Hispanics compared to the 2008 census numbers for registered voters, but they may have showed up in the category of 'Other' so that's not too bad either.
There have been two polls of registered voters in the past few months. One, from PPP, showed Romney ahead 50-43. (These results are not significantly different.) Another, from the University of Texas at Austin and the Texas Tribune, showed Romney ahead 46-38 among registered voters. There are six more, older polls from these two pollsters testing Obama vs Romney, with margins among registered voters of Romney +1 (UT), +2 (UT), +6 (PPP), +7 (PPP), +7(PPP), and +8(PPP). UT's 2008 polls showed McCain +11 among registered voters; final result: McCain +12.
What about other states near Texas - OK, AR, LA, MS, AL, KY, TN - shouldn't they show something similar? This gets tricky. The number of respondents in neighboring states in the Daily Kos poll is small - about 100-300. But, in every state, the margin between Obama and Romney is smaller than the margin between Obama and McCain. The average margin shift is 14 points in Obama's favor. Now, even if Obama were doing exactly as well as 2008, we should be seeing an average shift from election results of around 3-8 points in Obama's favor (see here for why), but 14 is a fair amount more than that. So neighboring states are behaving in a similar fashion to Texas.
Conclusion: This polling aggregation is likely too favorable for Obama, but it is not way off base.
So then, what's going on in the crosstabs to make Texas appear competitive?
Independents Don't Like Romney.
When I first saw the toplines, I thought it might be because Texas didn't have its primary until May 29th, midway through the polling period. Maybe grumpy Santorum partisans were skewing the results. But it turns out, not so much - those who hold Santorum dear to their hearts may not like Romney, but 90% of Texas Republicans say they'll vote for Romney in November even though only 69% voted for him in the primary.
No, the 47-47 tie is not because of cranky Republicans, it's because of a shift in Independents. More specifically, white Independents.
Only 32% of Independents have a favorable opinion of Romney. Meanwhile, 43% have a favorable opinion of Obama. Obama ties Romney among Independents, whereas in 2008, McCain won Texas Independents 62-36.
But among white independents (about 20% of the voters), McCain beat Obama 72-27. In this polling aggregation, Obama beats Romney 47-45.
I guess Romney isn't mavericky enough.
Other demographics with notable improvements for Obama are whites, men, moderates, and those with income >$100,000.
Republicans are not uniformly happy campers, with only 74% having a favorable opinion of Romney. However, 91% have an unfavorable opinion of Obama, so 90% are voting for Romney. Among Democrats, 81% have a favorable opinion of Obama, while 83% have an unfavorable opinion of Romney; 87% are voting for Obama. This is an enthusiasm gap in favor of Democrats. Click the following links for demographics and selected crosstabs for Obama versus Romney, Romney favorables, and Obama favorables.
I spent a little time looking at what the numbers would have to be for Obama to actually win Texas. I constructed a few charts showing percent Obama as a function of Obama's support among whites and Latinos for four scenarios. This assumes that support in other racial categories is the same as 2008.
1. Registration and turnout are the same as 2008, updated for 2012 demographics (esitmated from 2010 and 2011 census data).
2. Romney continues to act like the rich snot he is, and picks Boring Non-Evangelical White Guy From Up North for running mate. Turnout among whites declines from 88% of registered voters to 75%, similar to 1996. Turnout for other racial demographics remains the same as 2008.
3. The Hispanic population gets excited by Obama's immigration policy, and votes in proportions similar to those seen in other states. Registration goes up to 65% of Citizen Voting Age Population, and 85% of those vote. Turnout for other racial demographics remains the same as 2008.
4. Scenarios 2 and 3 combined. (Dream on!)
There's one big practical take-home lesson from looking at these scenarios that has been pointed out many times before: The Hispanic community has great potential power in Texas. If turnout were more like California, Texas would have a 6-point margin even without any change in vote share from 2008. Is it even possible for voting behavior to change so quickly? Probably not over one cycle, but California and New York did have large jumps in participation from 2004 to 2008, and Nevada has seen a huge change since 1996. Texas, meanwhile, has been lagging:
Now, for the scenario simulations. Below, you can see the results. The ranges start with Obama's 2008 vote share and increase by ten points. Decreasing Obama's vote share is not shown, because it's boring - Romney always wins - that doesn't mean it can't happen, of course. I assumed 2% of voters would vote for a third-party candidate, so 49% is a tie. Pink spaces have a margin of Romney +3 or more; blue spaces are Obama +3 or more.
For reference, Obama was polling better this year nationwide than he did in 2008 in the Hispanic community, prior to his immigration announcement. That announcement greatly increased enthusiasm in the community. An increase from 63% support to 73% support actually is not out of the question. Polling for Obama among Texas whites has ranged from 25%-36%.
Update: Title and text changed to make clear this is an analysis of an aggregation of multiple polls, and text added to emphasize that this is not likely to reflect November results.