Analysis: Don’t be fooled by high early US voting numbers

Early vote totals are historically not a good predictor of the outcome of the presidential race.

Voters in Lakeland, Florida after casting their ballots early, October 19, 2020 [Chris O'Meara/AP]

This is the point in a presidential campaign where people want to know, “Who’s going to win?” Rarely is there a certain answer to that question at this juncture, yet many still look for facts and figures to try to solve that mystery ahead of time.

One of those tangible pieces of evidence that has received a lot of pre-election publicity is early vote totals. Currently, more than 29.6 million people have voted early by mail or in person, shattering 2016 records.

If recent election analysis is prologue, we are likely to see pundits looking at these early statistics as a crystal ball regarding the final vote count. The problem is this data is an extremely imperfect prediction tool.

To wit: A huge story just prior to the 2016 election was a supposed “surge” in Latino turnout based on limited data that early voting offered. The New York Times confidently wrote just before Election Day: “Early voting data unequivocally indicates that Hillary Clinton will benefit from a long-awaited surge in Hispanic turnout, vastly exceeding the Hispanic turnout from four years ago.”

The reality wasn’t the bombshell it was made out to be, as the Pew Research Center summarised: “Overall turnout remained flat despite expectations heading into Election Day of a long-awaited, historic surge in Latino voters.”

This is just one example of the trap some analysts fell into by reading too much into early voting statistics in the last presidential race.

Traditionally, early vote totals are a minority of the total vote. In 2016, according to the US Election Assistance Commission, 41 percent of the total votes were cast early in person or by mail. And although that percentage is expected to be significantly eclipsed this year due to the pandemic, there’s still only a certain amount of information – albeit important information – that can be gleaned from early voting statistics.

  • In the 16 states that offer party registration data, we can tell how many voters by party affiliation have requested and returned ballots
  • In addition, nine states offer statistics on how many voters by party voted early in-person
  • There are even some states that break down early vote data by race and ethnicity

This is all interesting information and absolutely offers excellent insight into who is most enthusiastic about voting. This is also data that campaigns can use to gauge where to stress their last-minute get-out-the-vote efforts and with whom to target those efforts.

What it doesn’t tell us is which candidate these voters picked or how the early numbers fit in the overall turnout picture after all votes are cast on Election Day.

Even with the pandemic leading to a skyrocketing number of votes before November 3, there is still likely to be a significant in-person turnout on Election Day, especially among Republican voters. And that makes it very difficult to use this information as a predictor of the outcome.

Recent polling suggests that Joe Biden will benefit from early voting by a wide margin.

An NPR/PBS NewsHour/Marist poll released last week revealed that one in five voters say they’ll vote early in person with an additional one in three saying they’ll vote by mail. In the poll, Biden leads by 35 points among the in-person early voters and by 50 points among the mail-in voters.

But there’s an important reminder in the NPR poll: 45 percent of respondents say they plan on voting in person on Election Day. Of that 45 percent, two-thirds are Trump supporters and Trump leads day-of voters by 27 points over Biden.

Day-of voters have messed up many analysts’ predictions in recent elections. In both 2012 and 2016, North Carolina was seen as moving in Democrats’ direction during the early voting period. Both Barack Obama and Hillary Clinton lost the state after all of the votes were counted. In 2016, Florida and Arizona were being discussed as potential wins for Clinton because of early Latino turnout. Trump won both.

Those should be lessons for those seeking to draw conclusions during the next couple of weeks. The number of expected Election Day voters is significant. In 2016, exit polling showed that Trump won over the 13 percent of voters nationwide who made up their minds in the final week of the election. Enormous caution should be taken before linking a potential final outcome to early vote totals.

Source: Al Jazeera