Two things appear to be true: Early voting results are looking good for Hillary Clinton, and the polls are tightening. Clinton is now only polling about 2 points above Donald Trump nationally, according to the RealClearPolitics average, a difference that has narrowed drastically from what was a 6-point lead just over a week ago.
At the same time, new early voting numbers out of Nevada and Ohio this week were a good sign for the Democratic candidate, as turnout seems to be bouncing back to 2012 levels. In Nevada, Clark County (home to Las Vegas) is proving to be a promising firewall for Clinton. And in Ohio, similarly, Democratic-leaning counties are beginning to pick up the pace. Currently, early voting is at 60 percent of 2012 — in other words, 40 percent of votes cast in 2012 are still "out there."
This discrepancy can be confusing; why are these two metrics are seemingly pointing in opposite directions? But according to early voting experts, there is a simple explanation: On a national scale, there is not that much of a discrepancy at all, and moreover, results from state-by-state early voting and national polling simply can’t be compared.
Brian Hamel, a researcher with the Early Voting Information Center, said early vote numbers are also tightening up at this point: Nationwide, about 7.4 million Democrats have voted and 6.4 million Republicans have voted, CNN reported. “That suggests a relatively close national race,” Hamel, a political science PhD student at Universisty of California, Los Angeles, said.
This doesn’t discount Clinton’s gains in early voting, however. On that level, early voting numbers and polling are essentially apples and oranges. “We should not be concerned if polls do not match up to early voting metrics, given that they measure fundamentally different things,” Hamel said.
Early voting and polling measure different things
Early voting and polling are two different metrics.
The primary difference is that early voting results capture trends in actual votes over several weeks, whereas polls are a snapshot in time.
It’s an important distinction to take into account, as voters tendencies, as reflected in the polls, may be more influenced by the news events — whether that’s the sexual assault allegations against Donald Trump or the FBI reopening an investigation into Hillary Clinton’s emails. Those shifts in opinion may not reflect overall voting habits, and they don’t necessarily capture the nature of an early voter.
Early voting is usually a good measure of two things: enthusiasm and a campaign’s ability to mobilize its electorate. Lots of early voters means lots of decided voters, not voters that are easily swayed by news events.
And while there doesn’t seem to be much enthusiasm for either candidate this year, there is some speculation that people voting against the opposing candidate rather than affirmatively voting for their candidate might bring people to the polls.
When the early voting periods were about to open, Michael McDonald, an early voting expert with University of Florida, told CNBC he was watching the Hispanic and rural white vote. "There is speculation that Trump's rhetoric could entice Hispanics to vote against him, and there is some evidence in polls that Hispanic voting enthusiasm is running higher than normal," he told CNBC.
The same theory can be applied to black voters, Paul Gronke, director of the Early Voting Information Center, said.
"Trump continues to make statements that alienate African-American voters, and he struggles to get beyond 2 percent in some state polls. Under that circumstance, and faced with a well-oiled Democratic [get-out-the-vote] machine, why would African-American voters wait?" Gronke told me in September, noting that Trump also doesn’t seem to have much of an early voter ballot-chasing operation.
Just as national polling doesn’t predict state-by-state Electoral College results, early voting doesn’t predict the Election Day results
It’s also important to note that early voting results are not a good indication of actual election results. There are a lot of limitations: States reporting early voting totals don’t always include all counties; some states require party registration, like North Carolina, while others, like Wisconsin, don’t; and numbers derived from party registration are fallible — nothing is obligating registered Republicans or Democrats to vote for their party’s nominee, for instance. Making comparisons across state lines is also difficult. Some states have weeks of early voting and many early voting locations, while others have limited opportunities to cast a ballot early.
University of Denver political scientist Seth Masket crunched the numbers for FiveThirtyEight on how well early voting numbers predicted the final tally in past elections. He simply concluded, “The relationship is positive, but it’s pretty noisy. In other words, knowing how a party is doing in early voting doesn’t tell you much about how it will do once all the votes are counted.” In fact, he found that looking at early voting numbers in 2012 would give you “wildly misleading” results:
Democrats maintained substantial leads among early voters in North Carolina, Louisiana and West Virginia, and were trailing by a relatively narrow margin in Oklahoma, but still lost those states when all the votes were counted. Republicans won early voters in Pennsylvania and Colorado but lost the final tallies there. Maryland was a safely Democratic state in 2012, but the 75 percent of the early vote that went the Democrats’ way was a far cry from the 63 percent of the total vote they won once voting was finished.
Rather, “early voting might be best interpreted as an indication of how campaigns are going to use their resources in the final weeks,” Hamel said — it’s a good measure of how campaigns need to mobilize their voters for Election Day.
This comes into play in states like Ohio, where “lower overall turnout and lower Democratic turnout in Cuyahoga [County, home to Cleveland] suggests that Election Day get-out-the-vote in Ohio is going be huge for the Clinton campaign.”
As Gronke explains, early votes have been “banked,” just as you bank your money — “it protects you against short-term fluctuations in the electoral market,” he said.
Polling is not “banked.”