Opinion - Politics

Why experts are getting presidential election polls wrong – again

Key Points
  • In reporting about the 2020 presidential contest, it's likely that at least some of those stories quote nationwide polls.
  • Hopefully most Americans have figured out that both the primary elections to choose a presidential nominee and the general election to choose the president are state-by-state contests.
  • But statewide polls are much less reliable than nationwide surveys.
  • There's a multi-faceted silver lining to all of this if we're all willing to admit the truth about this polling conundrum.
A sign guides voters on Election Day.
Stephen Maturen | Getty Images

One year from Election Day 2020, America is still running the risk of being blind and deaf when it comes to several key aspects of our presidential election process. 

In reporting about the 2020 presidential contest, it's likely that at least some of those stories quote nationwide polls.

Yes, there are some reports on the statewide contests between the Democrats and versus President Trump. But the lion's share focus on the national picture.

That's what we saw Tuesday, with the major headlines coming out of the ABC News/Washington Post poll showing President Trump trailing five different Democratic candidates nationally. Publicizing polls like that may sound innocent enough, but here's the problem: that's not how we play this game.

Hopefully most Americans have figured out that both the primary elections to choose a presidential nominee and the general election to choose the president are state-by-state contests.

Yet most stories about the Democratic primary race focus on Joe Biden's enduring lead in national polls and not on how the Democrats are faring in early voting states like Iowa and New Hampshire.

It's as if we're intentionally blinding ourselves to the most pertinent facts every time anyone talks about nationwide polls.

The simple solution to all of this undue focus on national polls is to simply focus more on the state-by-state polls, right? That's what The New York Times did earlier this week with a special focus on the polls in six battleground states.

That makes sense in theory, but presents a new problem: statewide polls are much less reliable than nationwide surveys. Americans found that out on Election Night in 2016, when polls predicting victories for Hillary Clinton in the key swing states of Michigan, Pennsylvania, and Wisconsin all turned out to be wrong.

On the one hand, American voters are blinded to important facts because of the dominating focus on national polls in presidential elections. On the other hand, doing the right thing by switching that focus to statewide polling would subject voters to data that's more likely to be incorrect. It's the blind leading the blind.

If you're looking for a definitive reason why those crucial swing state polls were wrong, good luck. In the year or so since the 2016 election, we've heard a number of explanations that either don't hold up to scrutiny or cannot be objectively proven.

Perhaps the best example of that is the early explanation promoted by some pollsters who said that the 2016 swing state polls were wrong because most of them did not accurately weight them based on the respondents' level of education.

But here's the problem with that theory: the few statewide polls that were weighted for education levels also got the actual election results wrong. In some cases, The New York Times reported they were even more off the mark than the non-education weighted polls.

The other prevailing explanations are hard to fix or even prove. One theory is that a large majority of undecided voters decided to vote for Trump at the last minute. Another is that Trump was and is supported by disaffected Americans who are very unlikely to respond to pollsters at all.

Either way, voters and pundits alike are still flying pretty blind when it comes to statewide polls in a primary and general election system that's determined by statewide results.

So on the one hand, American voters are blinded to more important facts because of the dominating focus on national polls in presidential elections. On the other hand, doing the right thing by switching that focus to statewide polling would subject voters to data that's more likely to be incorrect. It's the blind leading the blind.

But other than that, everything's fine.

Actually, there's a multi-faceted silver lining to all of this if we're all willing to admit the truth about this polling conundrum.

First, voters could definitely do with more focus on what candidates are saying about the issues rather than the "horserace" aspect of our elections.

Many of the very same journalists who have been guilty of making the bulk of election "reporting" simply a series of repeating poll results have started to at least realize this is a problem for the health of our democracy and their profession.

Second, candidates who become more sensitive to the likely inaccuracy of statewide polling should become more likely to visit more of those states more often.

One of the great lessons of Hillary Clinton's failed 2016 bid was the fact that her decision not to make more frequent visits to key Rustbelt states came back to burn her.

It wasn't that the Clinton campaign didn't think visiting battleground states was important, it's just that it clearly didn't think Wisconsin, Pennsylvania, and Michigan were battleground states at all.

With all we now know about the lack of reliability of state polls, the lesson major campaigns should now learn is that almost every state is a potential battleground.

That's why Clinton's defeat should have led to a much better informed political class that chose to eschew conspiracy theory explanations for the 2016 results, rely less on polls, and focus on how to better listen to and connect with more voters.

When Richard Nixon made a big point of visiting all 50 states in the 1960 election, his eventual loss in that contest taught political pundits that the better way to win the White House was to game the system and only focus on battleground states. But that's led to a disconnect between politicians and many voters ever since.

The reality that every statewide poll is likely to be inaccurate and the additional reality that we don't even know why should do a lot to wipe out that disconnect. At the very least, it should open everyone's ears to a few more voices a lot more often.

Jake Novak is a political and economic analyst at Jake Novak News and former CNBC TV producer. You can follow him on Twitter @jakejakeny.