Winston Churchill once said, “Politics is not a game. It is an earnest business.” Yes, it is and never more so than in the current “worst of times,” with the world plagued by pandemic and the country torn apart by violence and hate. 

Issues and actions taken by our leaders at every level and by every party are no longer judged on their merits or wisdom or efficacy but whether they help or hurt Donald Trump. Emotion-based decision-making at this extreme doesn’t make for good politics, or governing for that matter.

Politics is serious business and necessary to how we run this country, but I have always looked at politics a little more clinically — numbers don’t lie. Well, accurate numbers don’t lie. So the spectacle in recent weeks with Democrats confidently predicting victory thanks to their presidential candidate’s big lead in the polls and the equally heated arguments by the Trump campaign that national polls don’t matter has added little to our understanding of what the polls mean and don’t mean at this stage of the race.

A little context is important here. Attend any campaign management training session during the last three decades, of either party, and you would have seen aspiring politicos taught that when it comes to winning a campaign, 50 percent plus one is the magic formula that delivers success in most races. And given that the two-party system still dominates in most places, it’s true. 

But where it isn’t always true is in the arena of presidential politics, where the Electoral College versus the popular vote comes into play, complicated by third-party candidates at times. As we saw in 2016, winning the popular vote doesn’t guarantee a candidate the White House.  And simply looking at the head-to-head numbers on one national poll or even a dozen doesn’t guarantee an accurate reading of presidential election outcomes.

3 percent’s a charm

But when looking at presidential polls, there is a number that matters — 3. I call it the 3 percent factor. When the race is at or under 3 percent, the head-to-head ballot test in national polls is not necessarily a clear predictor of who will ultimately win the presidency, and state polls are probably a better indicator of the race. But history tells us that when one candidate’s lead goes above 3 percent nationally, the size of the lead is likely to produce both at least a plurality in the popular vote and a win in the Electoral College. 

In 2016, Hillary Clinton won the popular vote by 2.1 percent. The RealClearPolitics average the weekend before the election had her ahead by 3.2 percent, within the margin of error. The polls accurately predicted her popular-vote victory, but it was state-by-state surveys that should have given the Clinton campaign and the media pause.

Because most national polls have a roughly a 3 percent margin of error, it’s also important to consider when analyzing surveys that ballot test leads could reflect plus or minus 3 points. So Clinton’s 3-point advantage just before the election, in reality, ranged from trailing by 3 points to a lead of 9 points.  Although there is a lot of statistical heft to how polls are conducted, they are not an exact science, but rather an inference.

On the other hand, election results are exact, and over time, they back up the theory that presidential races coming in at 3 percent or higher in the popular vote translate into Electoral College wins of 60 percent or more, with only three exceptions: in 1856 and 1860, when James Buchanan and Abraham Lincoln respectively took 59 percent of the Electoral College vote and in 1916, when Woodrow Wilson came in at 52 percent.

Starting with 1856, there have been 41 presidential elections.  Ten of those have been at 3 percent or closer in the popular vote. Of those 10, only four candidates were able to win the Electoral College but failed to win the popular vote — Rutherford Hayes (1876) Benjamin Harrison (1888), George W. Bush (2000) and Trump (2016).

Third-party candidates have occasionally played a role in election outcomes. In 1860, Lincoln won in a four-way contest. In 1912, it was a three-way race with Teddy Roosevelt running on the Progressive Party ticket and William Howard Taft as the Republican nominee. But it was Democrat Wilson who won the presidency by winning the 82 percent of the Electoral College vote with only 42 percent of the popular vote.

Later in the century, Strom Thurmond and George Wallace were players on the presidential stage, and in 2000, the race was so close that Ralph Nader’s candidacy likely cost Democrat Al Gore the presidency.

But whatever the dynamics at play, when the margin is beyond 3 percent, the winning coalition is big enough to win the Electoral College, clearly the assumption driving the Clinton campaign in 2016. When it’s at or under 3 percent, it becomes a state-by-state election, which explains the successful Trump strategy.

Survey overload

So what’s a voter to believe when it comes to polls? These days, it feels like there’s a survey released every hour on the hour with differing results. To fully understand any survey, we need to fully understand the survey methodology. How did it weight key demographics, including party ID, ideology, age, gender, race and education, to name a few. And what kind of voter screen was used to assess voter participation? 

The answers to these questions are crucial to determine the validity of the survey results. Any survey that fails to disclose the sampling and weighting parameters should be viewed with skepticism, especially when outliers tend to generate more clicks and eyeballs.

So with all this in mind, where is the 2020 race today? As of Monday, the RealClearPolitics ballot test average showed, Biden over Trump, 49 percent to 41 percent. Given a 3-point margin of error, it’s statistically possible that the contest at this moment could range from favoring Biden by a 52 percent to 38 percent margin to a much closer 46 percent to 44 percent Biden lead.

The challenge for the Trump campaign as time grows short is to tighten this race, targeting that all-important 3 percent. A state-by-state approach has some merit, but given Biden’s apparent lead, this will take a broader effort across states, reaching out to independents and women in particular to rebuild the coalition that elected him four years ago. With a pandemic and economic and social upheaval to deal with, this is about as challenging as it gets for any president.

David Winston is the president of The Winston Group and a longtime adviser to congressional Republicans. He previously served as the director of planning for Speaker Newt Gingrich. He advises Fortune 100 companies, foundations, and nonprofit organizations on strategic planning and public policy issues, and is an election analyst for CBS News.