The USC Daybreak presidential poll
Biden leads by 10.9 points
Updated
The USC Dornsife Presidential Poll is back for the 2020 election, offering insight into how voter’s attitudes change as election day approaches. Nicknamed the Daybreak poll because it updates each day just after midnight, this poll of more than 6,000 respondents predicts how the popular vote will be split.
Predicted outcome of the popular vote
The Daybreak poll works differently from other election polls. The project is led by USC’s Center for Economic and Social Research along with the school’s Center for the Political Future, in collaboration with researchers from the Santa Fe Institute and MIT.
It tracks the same panel of eligible voters for months leading up to the election, instead of drawing new respondents every time. Rather than simply asking people which candidate they expect to vote for, the poll asks each respondent to rate their probability of voting for each candidate as well as their probability of voting at all. That provides more information from each voter, which, in theory, should improve the accuracy of the results. The responses from each voter are then weighted, so that the result is more reflective of the overall voting demographic.
Each voter’s attitude toward the candidates is recorded twice a month. To ensure that all respondents are included in each day’s results, numbers are released as a fourteen-day average.
Divided results by party and place
The Daybreak poll offers more than just the overall results. It also separates Republicans and Democrats, and people who live in cities from those in rural areas. The results can differ, sometimes drastically.
Biden leads overall, with Democrats and in cities
Trump leads with Republicans and in rural areas
The race is close among independents and other areas
Want more politics?
Get the best Los Angeles Times election coverage. Sign up for the Essential Politics newsletter.
Why your friends could help predict the election
In addition to asking how panelists will vote, the Daybreak poll does something more. It also asks panelists to predict how their friends, family, and colleagues will vote.
This is a new method in public opinion polling that has yielded promising results. During the 2016 U.S. and 2017 French elections, the technique came closer to predicting the popular vote than traditional surveys which only asked respondents about themselves.
One reason for this survey’s success has to do with sample size. By asking about a voter’s social circle, researchers implicitly gain information on a larger, more diverse group of people, according to Mirta Galesic and Henrik Olsson from the Santa Fe Institute.
Thus far, the Daybreak social survey predicts a slightly closer race than the overall poll, though Biden still holds the lead.
Predicted popular vote, based on social circle questions
Experts who study this method have also attributed its efficacy to the fact that respondents sometimes feel more comfortable speaking truthfully about others than they do about their own private opinions.
The "Shy Voter" theory holds that some Trump supporters might fear harassment or feel embarrassed about publicly declaring their voting intentions.
This survey method also reveals how many American social circles have sorted into opposing political camps. Democrats and Republicans alike predict their social circles will vote for each party’s nominee. "Perhaps more surprising," wrote Galesic and Olsson, "is that both Democrats and Republicans seem to know that they are surrounded by like-minded people."
How people predict their social circle will vote
Members of both parties adjust expectations when asked to predict how their state will vote. Both Republican and Democratic panelists anticipated a much closer race among a group that expands far beyond their social circles.
How people predict their state will vote
Democrats, at least so far, have shifted more than Republicans when asked to make a statewide projection. Some experts attribute this to what they call the "Magic Trump" effect, which describes fears among Biden supporters that last-minute events could give Trump the win.
What happened in 2016?
In 2016, the Daybreak poll captured national attention as one of the few to show Trump leading in the popular vote throughout the race. Trump ultimately lost the popular vote to Democratic candidate Hillary Clinton, but went on to win the electoral vote.
According to Jill Darling, the survey’s director, an overrepresentation of rural voters led to an artificial bump for Trump. This year’s poll was adjusted to bring urban and rural representation into alignment with population percentages, Darling said.
How close were polls to predicting the 2016 popular vote?
Clinton popular vote margin
+2 percentage points
Points difference
USC Dornsife/Los Angeles Times Daybreak
5
4
IBD/TIPP
3
Marist College
Selzer & Company
1
ABC News/
Washington Post
2
CBS News/
New York Times
2
Monmouth University
4
+5
0
+5
Clinton popular vote margin
+2 percentage points
Points difference
5
USC Dornsife/Los Angeles Times Daybreak
IBD/TIPP
4
3
Marist College
1
Selzer & Company
2
ABC News/Washington Post
2
CBS News/New York Times
4
Monmouth University
+5
0
+5