So many polls. So many varying results.
Could President Donald Trump win Michigan a year from today? Or could anybody with a “D” after her or his name pound him by double digits?
It depends on the poll and that’s down-right frustrating for everybody.
Emerson College reported Monday that 57% of the 1,051 Michigan voters they asked said they preferred Bernie Sanders to Trump. The other 43% said Trump. The New York Times’ poll with Siena College has Sanders only edging Trump by two percentage points in Michigan (45% to 43%), well within the 5% margin of error.
Emerson College has Elizabeth Warren up 54% to 46% over Trump in Michigan, outside of the margin of error. The New York Times has Trump winning 46% to 39%.
Joe Biden v. Trump? Emerson College has it 56% to 44% Biden. The New York Times has it 45%-44% Trump, basically a tie.
This isn’t the only example. Look at the presumed 2020 U.S. Senate race between U.S. Sen. Gary Peters and Republican John James.
On Sept. 27, a Target Insyght poll commissioned by MIRS News had Peters up 53% to 37%, a 16-percentage point spread. A couple weeks later, the Marketing Resource Group had Peters only up on James by three percentage points (43% to 40%). Then Vanguard Public Affairs/Public Sector Consulting poll with Denno Research had it at a tie (39.5% to 39.3%).
And then Monday, Emerson College had Peters up 6 points (46% to 40%). Other polling not released publicly have Peters up by at least 10 percentage points.
This race hasn’t changed in the last six weeks. Why are the results so different?
It all starts with the sample. Who is going to show up on Election Day?
Before political pollsters ask any questions, they ask themselves: How many women will vote? How many men? African Americans? Latino? Rich people? Poor people? Liberals? Rural people? Young people? Older people? People who traditionally don’t vote? It’s from this demographic sample pollster pick the likely voters they call with their questions.
Pollsters look at history to compile this calculation and that’s what happened on Election Day 2016.
Based on a historic voting population and mix, Hillary Clinton should have beaten Donald Trump in Michigan. Hillary Clinton should have beaten Bernie Sanders in Michigan, too.
But 2016 didn’t have a typical voting mix. Predominately lower income, rural folks who hadn’t voted in years were inspired by Trump’s plain-spoken patriotism and showed up to voting booths in droves.
Those like John Yob’s Strategic National or Target Insyght through its exit polling accurately caught these people and dubbed the race a toss-up.
Are these people going to show up in 2020? Based on Trump’s latest choices of words — his graphic dramatization of the killing of ISIS leader Abu Bakr al-Baghdadi, for example — shows that’s what he’s going for. He wants to inspire even more of them.
Democrats are going for a larger voting sample, too. Over the weekend, the Michigan Democratic Party chairwoman and the Democratic interest group Priorities USA spent resources driving turnout for Tuesday’s municipal elections.
They weren’t pushing a particular candidate. They just want to get younger, female, minority voters in the habit of voting so they’ll do it Nov. 3, 2020.
Ingham County Commissioner Mark Grebner of Practical Political Consultants believes there’s going to be interest in the 2020 election regardless. A lot of interest. A year after correctly predicting the 2018 turnout at 4.25 million voters, he now says an historic 6 million Michigan voters will show up next year.
In 2016, roughly 4.75 million showed up and that was a huge turnout. But 6 million people voting in Michigan?
If Grebner is correct, the mix of voters isn’t going to be like 2016. It won’t be like 2018. It’ll likely be something we’ve never seen before.
Who are these people going to be? Until pollsters agree to that question, the polls will continue be all over the place.
(Kyle Melinn of the Capitol news service MIRS is at firstname.lastname@example.org.)