Barack Obama has emerged from the nominating conventions in his best position against Mitt Romney since spring, a 50-44 percent race among registered voters in the latest ABC News/Washington Post poll. But Romney recovers to a virtual dead heat among those most likely to vote, keeping the contest between them wide open.
Obama is the greater beneficiary of the back-to-back nominating conventions. For the first time he's numerically ahead of Romney in trust to handle the economy, the key issue of the 2012 contest, albeit by a scant 47-45 percent. Obama's seized a 15-point lead in trust to advance the interests of the middle class. And strong enthusiasm among his supporters is up by 8 points from its pre-convention level; Obama now leads Romney by 10 points in "very" enthusiastic support.
The 50-44 percent race among registered voters compares with a 46-47 percent Obama-Romney contest immediately before the conventions; while those shifts are within the survey's margin of sampling error, Obama is at his best vs. Romney since an ABC/Post poll in early April. That's the case even though fewer than half, 48 percent, approve of Obama's job performance in this poll, produced for ABC by Langer Research Associates.
The main change has been a shift among Democrats, coalescing around their party's nominee. Obama's support from Democrats who are registered to vote has advanced by 8 percentage points since before the conventions, to a near-unanimous 91 percent, matching his best; the number defecting to Romney has dropped by 6 points, to a mere 5 percent. Among other groups, Obama's support has reached a new high among men, while Romney is at new lows among moderates, whites and higher-income voters, all in ABC/Post polls since April 2011.
Additionally, there's been a shift in preferences in the eight tossup states identified by the ABC News Political Unit: Registered voters in these states now favor Obama over Romney by 54-40 percent, vs. 42-48 percent in these same states before the party conventions. And in the states with mid-levels of unemployment, it's 51-43 percent, vs. 40-53 percent pre-convention, further suggesting some progress for Obama in his economic arguments.
As noted, though, among likely voters - people who say they're both registered and certain to vote - the race squeezes shut at 49-48 percent, Obama-Romney, essentially unchanged since before the conventions (+2 Romney then, +1 Obama now, well within sampling error.) That means that Romney's supporters express greater intention to vote - a challenge for Obama's ground game, and a suggestion that the race could come down to turnout.
Obama faces another reality: No incumbent with an approval rating below 50 percent in September of an election year has been re-elected in ABC/Post polls dating to the Reagan presidency. However, one came close: Not in September, but in early August 2004, George W. Bush had just 48 percent approval among registered voters. That went to 52 percent the next month, en route to his re-election. (Among other presidents, it seems that only Harry Truman won re-election with less than majority approval as the election approached, but the only pre-election data point available is a Gallup poll from late June 1948, showing 40 percent approval.)
Romney has his own challenges; beyond his lack of traction on the economy, he's broadly seen as having failed to provide specifics of his governing plan - in effect a negative assessment of his convention presentation. Registered voters by 63-31 percent say Romney has not provided enough details on the policies he'd pursue as president. They divide much more evenly, 46-49 percent, on whether Obama has or hasn't given enough details on what he'd do in a second term.
Other results suggest opportunities for Romney. The "build that" theme may have legs; Romney is far more apt than Obama to be seen as understanding what it takes to build a successful small business, and registered voters by 53-35 percent think government programs make it harder, not easier, for small businesses to succeed - a position the opposite of what Obama has expressed. At the same time, Obama and Romney run evenly in trust to support small businesses, suggesting that Romney has yet to capitalize on this issue.
More broadly, registered voters by a 13-point margin, 53-40 percent, say government programs do more to interfere with people's lives than to improve them, a position again more in tune with Romney's image as an advocate of smaller government than with Obama's.
ATTRIBUTES - Obama's advantages, in turn, include a persistent lead over Romney in empathy; registered voters by 50-40 percent think Obama better understands the economic problems people are having, and continue to rate him as more personally likeable, by a broad and steady 61-27 percent. (When the two views are tested against each other, empathy independently predicts vote preferences to a far greater degree than does likeability.)
Obama is at his best against Romney in another attribute, being seen as the stronger leader, 50-42 percent; and runs numerically ahead, albeit not significantly, in being better able to work with both sides in Congress, 46-41 percent.
On a personal level, building on his advantage in likeability, registered voters by wide margins would prefer to have Obama to dinner at their home, think he is more likely than Romney to be "a loyal friend," and would rather have Obama care for them if they were sick.
On an attribute related more to crisis management than to personality, however, voters divide much more closely on who they'd rather have as the captain of a ship in a storm - Obama, 46 percent, or Romney, 43 percent. And this measure more strongly predicts vote preference.
In a more general question on political values, registered voters by 65-23 percent say it's more important that they trust what a candidate says than that they agree with that candidate. And trust in what both candidates are saying is weak, but better for Obama: Registered voters by 49-42 percent say his campaign is saying things it believes to be true, rather than intentionally trying to mislead people. On Romney these numbers go negative, albeit not significantly, 43-48 percent.
ISSUES - Among issues, the economy reigns, and with 53 percent of registered voters disapproving of how Obama's handled it, Romney should have chances. Barely a third say the country is better off than it was when Obama took office, and 38 percent think it would be better if Romney had been in charge.
Still, that's not a solid breakthrough. While 43 percent say the economy's gotten worse under Obama's presidency, most, 57 percent, don't think it would have done any better under Romney.
And while 20 percent say they personally have gotten better off under Obama, essentially no more, 24 percent, think they'd have done better under a Romney presidency.
Obama, moreover, has newfound competitiveness on related issues - for the first time running about evenly with Romney in trust to handle the deficit, and scoring 50-43 percent against him in trust to handle taxes - not a statistically significant margin given the sample size, but still a slight improvement, and Obama's best numerically this year. In general, it's a problem for Republicans when a Democrat is competitive on taxes and the deficit.
Beyond the economy, Obama has regained a significant, 11-point advantage over Romney in trust to handle terrorism, up from a scant 4-point gap in the spring. Obama has a 13-point lead in trust to handle international issues, an 11-point lead on social issues such as abortion and gay marriage and his widest advantage, 21 points, in trust to address women's issues. (Obama's lead on women's issues is similar among women and men alike). The candidates are rated more closely on health care generally and Medicare specifically, but the latter is an issue on which Obama's moved into a numerical edge, if not a significant one.
GROUPS - As noted, Obama's support among men who are registered to vote is his highest of this election, 48 percent; and among women, at 51 percent, his best since May; the gender gap is its smallest in more than a year. (Unusually, Obama's support is numerically higher among unmarried men than it is among unmarried women, the latter a core Democratic support group.)
Obama has his highest support among whites since April; 41 percent, while Romney's 52 percent in this group is a new low for him. And the contest stands at 47-45 percent among registered voters with household incomes more than $100,000 a year, Obama's first numerical advantage, and Romney's again lowest support to date, in this group. Further, Obama now leads by 56-39 percent among registered voters who've done postgraduate work, compared with a 47-49 percent contest in this group before the conventions.
Vote preferences among liberals remain similar to pre-convention levels (84 percent for Obama); but he's doing better with moderates, 56-33 percent, Obama's best in this group since May and a new low for Romney. Romney, additionally, has slipped to 73 percent support among conservatives, numerically his lowest since February; Obama's 25 percent support among conservatives is his best since February, and up 9 points from just before the conventions.
As mentioned, Obama has improved to a near-unanimous 91 percent support among Democrats, up from 83 percent before the conventions and matching his best; that occurred chiefly among Democratic men, who also moved in their preference for Obama vs. Romney on the economy.
Romney, for his part, has a similar 89 percent support among Republicans, essentially unchanged from two weeks ago. The two run evenly among independents, 46-48 percent, Obama-Romney, similar to the ABC/Post pre-convention poll.
Thirty-two percent of registered voters in this poll identify themselves as Democrats, 26 percent as Republicans and 37 percent as independents, continuing a record four-year preponderance of independents in partisan preferences. The split is almost identical among likely voters, 33-27-36 percent, Democrats-Republicans-independents. Democrats outnumbered Republicans by 7 points in the 2008 election.
Among reasons for the closer race between likely voters vs. all registered voters is that Obama does better among people with lower household incomes, $50,000 or less, and they're less likely to say they're certain they'll vote. Certainty to vote also is lower in related groups, including unmarried and younger adults, and racial minorities. And Romney does a bit better among independents who are likely to vote, with 54 percent support, vs. 48 percent among all registered voters.
OTHERS - Among other results, few registered voters - 18 percent - think the Romney campaign is using the issue of race in the election; most of them, 15 percent overall, think it's doing so inappropriately. More, 31 percent, think the Obama campaign is using the issue of race - 20 percent, inappropriately.
There are substantial differences across racial groups. Nonwhites are three times as apt as whites to think Romney's campaign is using race as an issue, and doing so inappropriately. On the other hand, whites and nonwhites in about equal numbers think Obama's campaign is using race as an issue - but whites are much more likely than nonwhites to think he's doing so inappropriately.
Separately, preferences overall continue to revolve largely around views of Obama - not surprising, since re-election campaigns tend to be a referendum on the incumbent. Three-quarters of Obama supporters say they're mainly "for" him, not mainly against Romney. Many fewer Romney voters, 45 percent, are mainly "for" Romney; half instead are chiefly against Obama. That said, affirmative support for Romney among his supporters is up by 8 points from July, a helpful trend for the challenger.
Finally, just 13 percent of registered voters say they might change their minds, down from 19 percent in July. But an indirect measure of movability - based on the anxiousness voters feel about their candidates and their interest in additional information - finds that more, 22 percent, remain persuadable, including about equal numbers of Obama and Romney supporters alike.
That result suggests that opportunity remains for both candidates to change the current dynamic. But the door may not stay open for long: at 32 percent, the number who are interested in more information about the candidates has dropped by 9 points from its pre-convention level.
METHODOLOGY - This ABC News/Washington Post poll was conducted by telephone Sept. 7-9, 2012, among a random national sample of 826 registered voters, including landline and cell-phone-only respondents. Results have a margin of sampling error of 4 points, including design effect. The survey was produced for ABC News by Langer Research Associates of New York, N.Y., with sampling, data collection and tabulation by Abt-SRBI of New York, N.Y.