×

Alabama's special election polls stink, and that's a good thing

  • There are massive discrepancies in the most recent polls tracking the Alabama U.S. Senate race between Republican Roy Moore and Democrat Doug Jones.
  • It's the latest proof that political polling is getting more inaccurate across the board.
  • The good news about bad polls is that it might push the voting public to pay less attention to them and more attention to the issues.

I have some bad news for all those who hoped the polling industry would improve after the disastrous results of the 2016 presidential election.

That bad news comes from Alabama.

It would be a gross understatement to say that the polls were "all over the place" leading up to Tuesday's special U.S. Senate election in Alabama. The last three published polls before the December 12 contest between Republican Roy Moore and Democrat Doug Jones were outrageously inconsistent.

Here they are:

Monmouth University poll: Moore 46 percent, Jones 46 percent (tie)

Emerson College poll: Moore 53 percent, Jones 44 percent (Moore +9)

FOX News poll: Jones 50 percent, Moore 40 percent (Jones +10)

Just think about the disparity of these polls. If either the FOX or the Emerson poll is accurate, then the other one will be 19 points off the mark! If the Monmouth poll is correct then both the FOX and the Emerson polls are ridiculously out of bounds.

There are a myriad of technical differences in the polls that may explain these discrepancies. Nate Silver at FiveThirtyEight has a more mathematical and data-oriented analysis of all that.

But the truth is polling is getting harder. The days when everyone was likely to answer their land line and/or spend a half hour with a live pollster coming to their doors are long gone. Cellphone users and their actual residencies are harder to nail down as more people move around and take their phone numbers with them. Voter turnout is harder to predict because it's much easier for a respondent to say they'll vote for a candidate than to actually turn out and do it.

All of the above challenges played major roles in the fatally flawed polling of the 2016 presidential election. The national popular vote polls weren't too far off, but we don't elect presidents based on the popular vote. The crucial statewide polls of states like Michigan, Wisconsin, and Pennsylvania were wrong and their 46 collective electoral votes made the difference in the election.

A year later, even Silver is baffled by the discrepancies in this Alabama race. For the record, he thinks Moore is still most likely to win but he's urging everyone to view every poll with extreme caution.

I've got a better idea. Let's start to ignore them as much as possible.

The reason isn't because the polling companies are necessarily politically biased. Some polling companies do work for specific political parties that pay for their services. But even they have the same important goal that transcends partisanship: They need to be right.

Remember, these polling companies are businesses that need paying clients or colleges and media companies that need the prestige that comes with being accurate. If they choose partisanship over accuracy they'll be out of business fast.

But the fact is less focus on the polls in every election would be a good thing for our democracy. Too much of what passes for political reporting is just the parroting of poll data. It's derisively called "horse race journalism" and it deserves the criticism. This is an especially big problem in the early stages of a campaign when a candidate's poll numbers get more coverage than his or her actual positions.

Focusing on polls also leads to an unfair incumbency advantage that comes from name recognition. A 2011 Princeton University study showed that simply knowing a candidate's name gives those candidates a boost in the polls that's hard to overcome.

As more and more of these polls come up short, everyone from journalists to voters may learn to stop relying on them so much. It's not a stretch to conclude that the results would be at least a little more voter engagement and turnout. When polls aren't available, voters are logically less likely to assume any election is a foregone conclusion. That assumption plays a massive role voter turnout.

A more honest and issue-based American electoral process would be a win for everyone. The only polls that should matter are the ones taken on Election Day at the actual ballot box.

Commentary by Jake Novak, CNBC.com senior columnist. Follow him on Twitter @jakejakeny.

For more insight from CNBC contributors, follow @CNBCopinion on Twitter.

WATCH: Moore rides a horse to polling place