WASHINGTON — Hillary Clinton heads into the final presidential debate with Donald Trump on Wednesday night with a lead in a new Bloomberg Politics poll and many trends in her favor.

“This poll shows movement toward Clinton with all the right groups it takes to win—including men and those without a college degree,” said Bloomberg pollster J. Ann Selzer. “Their alignment with Clinton is a formidable change in the algebra.”

Among the key findings from the national survey, released Wednesday morning:

• In a two-way race, Clinton leads 50 percent to 41 percent.

• The margin is the same when third-party candidates are included.

• Clinton is cutting into Trump’s base of voters who are white with no college degree. She is ahead 48 percent to 44 percent among those with no college degree. With college-educated whites, Clinton is ahead by 13 points.

From Bloomberg’s John McCormick story: “Support for Trump among critical groups of voters, including men and the less educated, has weakened in the campaign’s closing days, a trajectory that could translate into a landslide loss for Republicans in the Electoral College and setbacks in down-ballot races that will determine control of the House and Senate.”

METHODOLOGY: “The Bloomberg Politics Poll, conducted Oct 14 – 17 for Bloomberg Politics by Selzer & Co. of Des Moines, IA., is based on interviews with 1,006 U.S. adults ages 18 or older who say they will definitely vote or have already voted in the 2016 general election.

“Interviewers with Quantel Research contacted 1,328 U.S. adults with randomly selected landline and ce ll phone numbers supplied by Survey Sampling International. Responses from the full probability sample were weighted by age, race, and education to reflect the general population based on recent census data. Interviews were administered in English.

“Percentages based on the subsample of 1,006 likely voters in the 2016 general election may have a maximum margin of error of plus or minus 3.1 percentage points,. This means that if this survey were repeated using the same questions and the same methodology, 19 times out of 20 times, the findings would not vary from the percentages shown here by more than plus or minus 3.1 percentage points. Results based on smaller samples of respondents — such as by gender or age — have a larger margin of error.”