Donald Trump has almost no plausible path to the White House unless he wins Florida, a rapidly changing state where Hispanic voters could deal a decisive blow to his chances.

But a new poll Monday, by The New York Times Upshot/Siena College, suggests that Trump is keeping his hopes alive in Florida, the largest and most diverse of the crucial battleground states. The reason: White voters favor him by a large margin.

Hillary Clinton leads by a single point, 41 to 40 percent, among likely voters in a four-way race that includes Gary Johnson and Jill Stein. The race is tied in the head-to-head race, 43-43 percent.

The poll, the first of its kind by The Upshot, was based on voter records that allow unusually detailed analysis of the electorate.

It indicates that Trump leads Clinton by 51 percent to 30 percent among white voters — and that includes all white voters, not just those who have been so vital to his campaign: whites without a college education. Clinton is winning white voters registered as Democrats by only 63 percent to 17 percent.

But she appears to have a huge edge among Florida’s Hispanic voters. She leads Trump by a 40-point margin, 61 percent to 21 percent, more than doubling the 18-point margin President Barack Obama recorded four years ago, according to Upshot estimates.

She is also doing very well among black voters, though not quite matching the huge margin or the enthusiasm that Obama enjoyed in 2012, at least not yet.

The story of demographics is playing out across the country. National polls suggest that the bottom has fallen out for Clinton among white voters without a degree, causing her lead in national surveys to narrow significantly.

Trump is leading in recent surveys of Ohio and Iowa — two states with a lot of white working-class voters whom Obama won fairly comfortably four years ago.

If Clinton continues to struggle among white voters nationwide, diverse states like Florida or North Carolina, where The Upshot will release a poll later this week, will become more important to her chances.

If she wins Florida, it will be extremely difficult for Trump to win the presidency. He would need to sweep the most hotly contested battlegrounds — Iowa, Ohio, Pennsylvania, North Carolina, Nevada and New Hampshire — then win somewhere Clinton is thought to have a considerable edge, like Michigan or Virginia.

The same trends are not so evident in the state’s Senate race. Marco Rubio, the Republican senator running for re-election, leads his Democratic challenger, Patrick Murphy, by 6 percentage points, 48 percent to 42 percent.

Trump’s unpopularity with nonwhite voters has not seemed to hurt Rubio’s chances. He trails among Hispanic voters by just 6 points.

Shifting state

Unlike many public polls, the Upshot/Siena survey was conducted using voter registration files, the core of the “big data” that has transformed campaigning over the last decade. The voter file data — from L2, a nonpartisan voter file vendor — includes information on the vote frequency, race and partisanship of every voter in the state, a big advantage for polling.

Our model suggests that the race has the potential to reshape the familiar political geography of Florida. Miami-Dade County, once fairly competitive, could be on the cusp of becoming a Democratic bastion.

Heavily Cuban enclaves in west Miami and Hialeah are divided, according to the model, even though they voted heavily for Mitt Romney in 2012.

The I-4 corridor looks like a patchwork of racially polarized Democratic and Republican enclaves. Many areas where Democrats used to be competitive with white voters — north of Tampa or around Daytona Beach, for example — appear to lean to Trump. It is gains like these that have helped Trump stay in the race, despite his loss of ground in South Florida.

There are growing Puerto Rican enclaves south of Orlando poised to vote overwhelmingly for Democrats.

And there are overlooked, rapidly growing and mostly white communities set to vote overwhelmingly for Trump. One such place is The Villages, a retirement community in Central Florida with a population now over 150,000. It was the fastest-growing city in the United States in 2013 and 2014, according to the census.

Clinton challenge

Clinton may have a narrow edge among likely voters, but the race is not quite so close among registered voters, who support her by a 4-point margin.

Her challenge is straightforward: to get less likely voters to the polls. Trump has a considerable lead among the likeliest voters, the older, generally whiter voters who regularly turn out in primaries and midterm elections. He has a 5-point lead, for instance, among voters who participated in the 2014 midterm election.

The presidential election will inevitably draw millions of additional voters from the pool of less regular voters, who are younger and more diverse. Clinton has a sizable lead among these less regular voters. The poll, for instance, gives Clinton a 10-point lead among registered voters who did not participate in the 2014 midterm elections.

But more than half of younger voters in Florida say they have an unfavorable view of her. And young voters are more likely to consider third-party candidates.

The potential upside for Clinton is obvious. If everyone in the state turned out and chose between one of the two major candidates, the model suggests, Clinton might lead by 6 points.

But these are not people with a great track record of voting. A lack of enthusiasm among younger voters wouldn’t just mean an older electorate; it might also mean a whiter electorate.

Overall, 69 percent of likely voters in the survey were non-Hispanic whites (as indicated on their voter registration form when they registered to vote), compared with 68 percent in the 2012 presidential election and 73 percent in the 2014 midterm electorate. The main reason for the slightly whiter electorate is a projected decline in the black share of the electorate.

Newly registered voters will probably reduce the white share of the electorate slightly.

Poll choices

All polls, of course, are subject to a margin of error and other potential sources of error. As a result, it is generally better to look at an average of recent surveys, which currently shows a very close race in Florida.

Decisions by pollsters matter, too. One such choice is over the likely-voter model, the process of determining which registered voters are likely to vote on Election Day.

Our likely-voter screen averages two methods: asking voters whether they’ll vote, and using a statistical model to estimate the probability that voters will participate in the election.

Clinton fared worse than she did among registered voters under both measures. But if we had used only self-reported vote intention, as many public polls do, Clinton would have had a 2-point lead. If we had used our model based on vote history, the race would have been tied.

With the result so close, there are different choices we could have made that could have given either Trump or Clinton the lead. In seven weeks, we will have a decisive answer.

About the poll

The New York Times Upshot/Siena College poll of 867 likely voters in Florida was conducted from Sept. 10-14.

The sample was selected from an L2 voter file stratified by age, region, race and a modeled turnout score. Voter records from each strata were selected in inverse proportion to the anticipated response rate for each strata, based on a June-July test.

Interviews were conducted on both landline and cellphones and in English and Spanish. Overall, 59 percent of interviews were completed on cellphones, and 4 percent were completed in Spanish. Interviewers asked for the person listed on the voter file; no interviews were attempted with other individuals available at the number.

The sample was balanced to match the demographic and political characteristics of active registered voters in the L2 voter file by age, race, gender, party registration, region and a modeled turnout score. The voter file data on respondents, not the self-reported information provided by respondents, was used for weighting.

Likely voters were determined by averaging a self-reported likely-voter screen and a modeled turnout score.

>> Self-reported likely voters were those who indicated that they were “almost certain” or “very likely” to vote, or rated their chance of voting as a “9” or “10” on a scale from 1 to 10.

>> The turnout score was based on a model of turnout in the 2012 presidential election. The probabilities were applied to 2016.

The probability that a registered voter would turn out was based on the average of whether they were a self-reported voter and their modeled turnout score.

© 2016 The New York Times Company