As the presidential race moves into a key two-week period, with the announcement of running mates and the party conventions, Donald Trump has taken an apparent slim lead over Hillary Clinton, based on strong support from white voters, particularly men.
That finding, from a USC Dornsife/Los Angeles Times Daybreak tracking poll, a new survey that begins publication Friday, marks a significant shift in a race that most polls indicated Clinton has led since mid-May.
It comes amid a flurry of other surveys, both nationally and in battleground states, that show support for the former secretary of State declining since last week when FBI Director James B. Comey characterized her handling of classified material while in that office as "extremely careless." Comey also said her conduct and that of her aides did not clearly violate the law or warrant prosecution.
What isn't known is whether the new surveys are capturing Clinton at a low that will prove temporary, as voters react to Comey's criticism and the renewed attention to her use of a private email server, or whether they reflect a more lasting shift that could hobble the presumed Democratic nominee for the remainder of the campaign.
The polls do not yet measure, for example, whether Clinton will receive a significant boost from Sen. Bernie Sanders' endorsement on Tuesday. The next couple of weeks also could prove pivotal as voters tune in to the campaigns during the conventions.
Trump continues to face formidable obstacles to winning. Even as new surveys show the race tightening, he has not significantly increased his support: Since February, when he began to dominate the Republican primaries, his backing in head-to-head matchups with Clinton has rarely risen above 40%.
Instead, several new surveys show Clinton's support declining, while the number of voters saying they will vote for a third-party candidate has risen.
In the new tracking poll, through Thursday night, Trump led Clinton 43% to 40%. That's within the poll's margin of error of 3 points in either direction, meaning the apparent lead could be the result of chance.
By Friday morning, the poll, which will be updated every day through the election, was showing a decline in Trump's lead.
The poll shows big gaps along the lines of race, gender, age and education that have surfaced consistently during the campaign. Through Thursday's results, Trump led among men, 47% to 36%, while Clinton had a smaller, 41%-34% edge among women. Trump led among voters 45 and older, Clinton among those younger.
Some of Trump's strongest support comes from white voters who have not graduated from college, among whom he led 53% to 24%. Clinton, by contrast, dominates among minorities, leading 77% to 3% among blacks and 51% to 30% among Latinos.
Clinton also held a narrow edge among white college graduates, 42% to 40%. If she wins that group, Clinton would be the first Democrat to carry white college graduates since polls began asking such demographic questions in the early 1950s.
The poll also offers some support for a prediction that Trump's backers have made – that he would appeal to disaffected voters who did not cast ballots in 2012. Those who did not vote that year or voted for a minor-party candidate were more likely to favor Trump than Clinton, the poll indicated.
Although respondents to the poll narrowly favor Trump, they don't necessarily expect him to win. In a separate question asking people who they think will prevail, Clinton led 53% to 41%.
Research has shown that that question often – although not always – forecasts election results more accurately than asking people their voting intention, particularly months before the vote is counted.
The Daybreak tracking poll differs from traditional polls in two major respects. Rather than questioning a different group of respondents for each poll, the survey relies on a panel, currently consisting of about 3,000 people recruited at random to represent U.S. households.
The panel is part of a larger Understanding America Study conducted by USC's Dornsife Center for Economic and Social Research. The election survey is being done in partnership with The Times and USC's Jesse M. Unruh Institute of Politics.
Because of the panel design, "we have the same people every time, so changes in the poll are really people changing their minds," rather than the result of variations in who answers a particular survey, said Arie Kapteyn, the director of the USC Dornsife center, who pioneered the approach for the 2012 election while at Rand Corp.
The panel design typically shows less volatility than traditional polls. Four years ago, it proved more accurate than most other surveys in forecasting the election result, although "maybe that was beginner's luck," Kapteyn said.
The other major difference is that the poll, using a 1-to-100 scale, asks respondents to say what the chance is that they will vote as well as the chance that they will cast a ballot for Clinton, for Trump or for another candidate. The results are weighted based on those probabilities, so that a voter who is 100% sure of his or her choice has more impact on the forecast than one who is 60% sure.
That approach is one way to resolve "one of the biggest problems that polls have – deciding who is going to vote," Kapteyn said.
Most polls use several questions to try to determine who is a likely voter and make a forecast based on that, but efforts to predict likely voting are often wrong, particularly far in advance of the election. Employing probabilities means "you get to use all the data," Kapteyn said. In theory, that should lead to more reliable results.
Several other polls released in the last couple of days point to the damage that the email issue has caused Clinton.
A New York Times/CBS News poll released Thursday morning, for example, showed that voters by a wide margin said Clinton would be better than Trump at handling several major issues.
But 67% of voters in the survey said Clinton was not trustworthy, an increase of 5 points from a CBS survey taken last month. The two candidates were tied in the new New York Times/CBS poll, with each receiving 40% of the vote.
A national poll by Marist College for McClatchy newspapers also showed a significant drop in support for Clinton. That July 13 survey showed Clinton leading Trump, 42% to 39%, but in April, a similar Marist poll had found Clinton with a 9-point lead.
Clinton's support has dropped 8 points since then. Trump dropped 2 points, and 13% now say they would not vote for either of the two, the poll found.
A 50-state compilation released Thursday by Morning Consult, a media and polling firm that conducts surveys online, found Clinton still leading in enough states to win the presidency, with 320 electoral votes for her, 212 for Trump, and Iowa's six electoral votes a dead heat.
In addition to Iowa, there were eight states where Clinton and Trump were within 2 points of each other, according to the firm's surveys, which were based on responses from 57,000 voters in all 50 states. Those were Ohio, where the firm found Trump narrowly ahead; and Florida, Georgia, Maine, Michigan, New Hampshire, Pennsylvania and Wisconsin, all of which leaned slightly to Clinton.
Compared with a previous round of surveys in April, Clinton had lost significant ground in Iowa, Wisconsin, New Hampshire and Ohio. She had gained, however, in Georgia and Nevada, both states with large minority populations. Nevada is now firmly on the Democratic side, the Morning Consult surveys showed.
New polling from Fox News showed Clinton leading in two other states where minority voters are likely to hold the key – Virginia and Colorado.
Additional evidence of greater risk for Clinton in battleground states came from three other surveys released Wednesday.
Wisconsin's closely followed Marquette Law School poll showed Clinton with a 4-point lead, 45% to 41% among voters likely to cast ballots in that state in November. That was down from a 9-point margin, 46% to 37%, among likely voters in the same survey last month.
Polls in Ohio, Pennsylvania and Iowa by Marist for NBC News and the Wall Street Journal found Clinton ahead in Iowa and Pennsylvania and tied with Trump in Ohio. In all three states, however, Clinton had lost ground since previous Marist polls.
Polling by Quinnipiac University of voters in Pennsylvania, Florida and Ohio similarly showed Clinton losing support. Their surveys found Trump narrowly ahead in Pennsylvania and Florida and a tie in Ohio.
The Quinnipiac polls have generally shown Clinton doing somewhat worse than other surveys, so what's most significant is the downward trend for her, which matches that found by Marist and Marquette.
For more on Politics and Policy, follow me @DavidLauter