Daybreak poll analyzes likely outcome of election


A poll conducted by the Jesse M. Unruh Institute of Politics in partnership with the Los Angeles Times is applying a new approach to tracking political opinions. The USC/Los Angeles Times “Daybreak” Poll implements a rating scale to ask a panel of approximately 3,000 participants questions about their outlooks on the presidential race.

The poll, which launched in July 2016 and will run through the November election, is part of the Understanding America Survey managed by the Dornsife Center for Economic and Social Research. The UAS is a panel at USC comprised of 5,500 individuals representing the United States. The study is an ‘Internet’ panel, meaning that participants use computers, smartphones and tablets to engage in the survey. Respondents answer questions about a variety of topics, including their household demographics, health, retirement plans and thoughts on the election.

The UAS election panel consists of around 3,000 randomly recruited participants from the wider panel who have agreed to answer questions about the election. Each day, around 400 individuals out of those 3,000 are asked to state the likelihood that they will vote in the election, the candidate that they will likely vote for and the chance they think each candidate has of winning. The results of the poll are weighted to take into account demographic characteristics, such as race, gender and income, from the U.S. Census Current Population, and adjusted to the 2012 presidential election outcome based on how participants voted back then. The results are regularly updated online, along with averages of all the prior week’s responses.

The Daybreak Poll exhibits a distinguishing characteristic from most polls, which will ask people which candidate they prefer and exclude people who are undecided from any analysis of candidate support. The Daybreak Poll, however, asks participants to rate, using a 0-100 scale, their chances of voting for Clinton, Trump or another candidate. Because of this, every participant will influence the results, which provides insight into how uncertain voters are currently feeling.

Unruh Institute Director Dan Schnur said that this method allows for better monitoring of how voters’ opinions are changing.

“Our poll isn’t simply measuring the preference for candidates, but rather the intensity of voter support,” Schnur said. “It’s become increasingly clear over the past few months that while Clinton enjoys support from a larger number of voters in this election, Trump supporters are much more fervent in their support of his candidacy.”

Some have expressed concerns that the Daybreak Poll shows a Republican bias in terms of candidate support, because of how it factors in the way people voted in the 2012 election. The poll weighs the sample of participants so that 25 percent of the sample are people who claimed to have voted for 2012 Republican party nominee Mitt Romney and 27 percent for President Barack Obama. But problems can occur with self-reported past voting when people aren’t honest about how they voted. Some polls have revealed that the percentage of people who — after an election — say that they voted for the winner exceeds the winner’s actual vote. This weighting could potentially cause too many Republican voters to be included the sample, giving Donald Trump a boost in support.

Jill Darling, the survey director of the UAS, said that all pollsters vary in terms of how they choose to collect and model their data. She doesn’t think that there is a definitively accurate technique to poll voters.

“One study found that people who had not voted for an incumbent were less likely to participate in a subsequent study about election results,” Darling said. “We can’t know if people are under-reporting votes for a non-incumbent, or whether the sample itself may just not have enough of those people [who didn’t vote for an incumbent] in it.”

Some USC students believe that while the poll is useful, it does have certain drawbacks. Ali Main, a junior studying broadcast journalism, said that although she consults the poll frequently, she doesn’t think it’s necessarily indicative of the outcome of the actual election.

“The Daybreak Poll is definitely interesting because it constantly surveys the same panel of voters,” Main said. “However, I think some of the polling choices, including weighing the sample based on people’s self-reported voting in 2012 and using the 0 to 100 rating scale and the effects that these choices have on the data prove that the results of any one poll are not enough to accurately predict what will happen on election day. I look at the results of this poll on a regular basis, but I can’t base my thoughts on the state of the election on polling data alone.”