USC Dornsife Daybreak Poll Final Full-Wave Estimate
The 2020 USC Dornsife Daybreak Poll tracked changes in Americans' opinions throughout the campaign for the White House from August 11 to November 2. A total of more than 6,400 eligible voters who are members of CESR’s Understanding America Study, a probability-based internet panel, participated at least once, with an average of about 5,300 participants per wave.
Tracking poll participants answered questions every other week about who they intended to vote for. Once a month they answered a longer poll which focused on what they care about most in the election, their attitude about the election, and their preferred candidates.
The tracking poll graphic was updated just after midnight every day of the week. Each day’s data point represented the estimate among voters over the previous fourteen days, closely representing the poll's full sample. The final estimate in the tracking graph showed a nearly 11 point Biden lead (53.5% to 42.6%). This estimate is wider than the 8.4 estimate among national polls in fivethirtyeight.com's national polling average.
The chart below illustrates a slightly different way to calculate the tracking poll data. Each data point is based on one of the poll’s six full waves. This method provides a final estimate of a nine percentage point lead for Biden on the day before the election. The difference in the two estimates is attributable to slightly different survey populations. We will use this method of estimating outcomes, rather than the "rolling" 14-day estimator, in our post-election analysis.
States are still counting votes at this time, but both these methods of calculating our estimate result in a wider margin for the national popular vote than what it's likely to be.
One other estimator on our tracking site – the social circle vote question - predicted a much closer race and may end up being the closest estimator of the four question types we tested this year.
Dornsife Daybreak Tracking Poll Estimate by Wave – probability vote
In addition to the tracking poll, we conducted three full length polls once a month during the pre-election season, and a post-election poll. Releases from these data collections are available on our 2020 presidential election data site, along with the tracking poll longitudinal file. Data from the pre- and post- election polls will be released on the data site in December.
The 2020 USC Dornsife Presidential Election Daybreak Poll tracked changes in Americans' opinions throughout a campaign for the White House from August 11 to November 2. A total of more than 6400 eligible voters who are members of CESR’s Understanding America Study, a probability-based internet panel, participated at least once, with an average of about 5300 participants per wave. Tracking poll participants answered questions every other week about who they intended to vote for. Once a month they answered a longer poll which focused on what they care about most in the election, and on their attitudes toward the election, and toward their preferred candidates. The "Daybreak poll" was updated just after midnight every day of the week. Each day’s data point represents the estimate among voters over the previous fourteen days, representing the poll's full sample. We also offer a 7-day graph which shows more variation due to it relying on a half sample. For more information about the poll and its available data and information, visit the survey methods tab.
The 2016 USC Dornsife / LA Times Presidential Election Poll followed the pioneering approach we first adopted in the RAND Continuous 2012 Presidential election poll, by tracking changes in Americans' opinions throughout the 2016 campaign for the White House. More than 4500 respondents who are members of our representative online panel were asked questions on their voting intentions, what they cared about most in the election, and on their attitudes toward their preferred candidates. The "2016 Daybreak poll" was updated just after midnight every day of the week.
Our final poll measured Donald Trump with a three point lead. This was widely misunderstood to mean that we were one of the few polls to predict Donald Trump’s win in the electoral college. In fact, Hillary Clinton was the winner of the national popular vote, by 2 percentage points.
As we had always intended, we revisited our experimental poll to consider what had worked well and what could be improved. Our examination of sampling and weighting methods revealed that we had an overrepresentation of rural voters in our sample, a possibility that we simply overlooked at the time, as rurality is not a typical weighting variable.
When we adjusted our poll according to census benchmarks for rural/urban population, our final estimate had Hillary Clinton over Donald Trump by a one point margin. We of course would have preferred to have that outcome on November 4, 2016, but we were pleased to find where our error lay and to be able to correct it. More information in our Q&A about this poll.
The poll was part of the ongoing Understanding America Study: (UAS) at the University of Southern California’s (USC) Dornsife Center for Economic and Social Research, It was conducted in partnership with the then Jesse M. Unruh Institute of Politics (now Center for the Political Future) and the Los Angeles Times.
Every day, we invited one-seventh of the members of the UAS election panel to answer three predictive questions: What is the percent chance that… (1) you will vote in the presidential election? (2) you will vote for Clinton, Trump, or someone else? and (3) Clinton, Trump or someone else will win? As their answers come in, we update the charts daily (just after midnight) with an average of all of the prior week’s responses. To find out more about what lies behind the vote, each week we also ask respondents one or two extra questions about their preferences and values. The team responsible for the USC Dornsife/LA Times Presidential Election Poll four years ago developed the successful RAND Continuous Presidential Election Poll, which was based on the same methodology. For more information, documents and data, please visit the Understanding America Study’s elections page.
Tracking Poll Design
On August 11, 2020, USC’s Center for Economic and Social Research (CESR) invited 8,355 eligible voters who are active members of CESR’s Understanding America Study (UAS) probability-based internet panel to participate in an ongoing election tracking survey. The baseline data and consent information was collected in UAS survey 306.
Each study member who agreed to participate was randomized to respond on a pre-assigned day of the week, distributed so that our full sample participates over a 14-day period. Respondents have until their next assigned wave day (or 14 days after their assigned date) to complete the survey. Data for the full sample is nearly complete after the first 14 days, but not final until the end of the full 28-day wave.
We provide four models of the presidential vote: probability-based personal voting, estimates based on expectations of how others in the respondent’s state are voting, expectations about how respondents’ social circle is voting, and a categorical vote estimation among a group of likely voters who are asked who they would vote for it the election were held today. The categorical vote question explicitly asks about the Democratic, Republican, Green and Libertarian candidates by name and party. All others ask about the two main candidates and "someone else". For information specific to this tracking poll including question text, survey design, weighting, and contact information, please consult our tracking poll methodology document.
The daily updated data points in the graphs on this site represent estimates of the national vote for Donald Trump (red line) and Joe Biden (blue line), using the four models. These are calculated based on the previous fourteen-day or seven-day cumulative responses to the tracking poll. The gray bars represent the margin of error for each estimate. Differences between the estimated vote for the two main candidates is not statistically significant if the lines are in the gray area. Graphics data in csv format may be downloaded from the context menu (three bars) in the upper right hand corner of each graph. Also available are images of the graph in various formats.
Information, Documentation and Data
Survey questionnaires, toplines, press releases, graphics, and data are available on our 2020 Election Data page, along with details about how we sample and weight the tracking poll. Check out our Q&A for information about this poll, and similarities and differences in methodology between 2016 and 2020. Microdata files corresponding to today's graphic data point, and the longitudinal tracking poll data file are also available to registered users on that site. For detailed information about the Understanding America Study online panel design, including sampling, recruitment, weighting, and attrition, please visit the methodology section of our UAS data pages.
Public Use Data files
Microdata files are available to registered UAS data users for download from our 2020 Election Data site. To register, visit our registration page.
The csv files listed below contain the results plotted in the charts, with some additional information, such as sample size.
Registered users of the Understanding America Study can download the underlying individual-level (micro) data from the UAS datapages. Registration is free. You can download the data user agreement here, complete and sign it, and return it as indicated on the form.
Click here for more information on these files.
The USC Center for Economic and Social Research's Dornsife Presidential Election Poll is part of the ongoing Understanding America Study (UAS) at the University of Southern California's (USC) Dornsife Center for Economic and Social Research (CESR). The project represents a pioneering approach to tracking changes in Americans' opinions throughout the period leading up to the election. Respondents in our representative panel of U.S. households are asked questions on a regular basis.
For detailed information about how the survey is conducted, including full question wording and weighting / estimation procedures, please visit our Survey Methods tab. For information about the Understanding America Study (UAS) web panel in USC's Dornsife Center for Economic and Social Research (CESR), including inquiries about conducting surveys with the panel, or to meet the Election poll team, or view our publicly available data sets, please visit the UAS site at https://uasdata.usc.edu or contact CESR Director Arie Kapteyn (firstname.lastname@example.org), UAS Survey Director Jill Darling (email@example.com) or CESR Managing Director Tania Gutsche (firstname.lastname@example.org). CESR offices are located on the University Park Campus of the University of Southern California in Los Angeles, California and in Washington, D.C.
For questions relating to the poll's content and findings, please contact UAS Survey Director Jill Darling at email@example.com.
The 2020 USC Dornsife Presidential Election Poll was initiated and financed by USC
Copyright © 2020 The University of Southern California. All Rights Reserved.