"The polls were very good, despite lots of concern about the polls," said Grossmann, the director of the Institute for Public Policy and Social Research (IPPSR) at Michigan State University. "They also did not match the previous years' bias in Michigan – so some sign that we might have at least escaped, at least in midterm elections, some of the problems we have with polling."
During the IPPSR's General Election recap on Tuesday, Grossmann said reductions in politically biased polling "is surprising" considering that response rates have been continuously dropping for telephone polls and online polls still obtaining unrepresentative baseline samples.
He shared a chart constructed by Alexander Agadjanian, a political science doctoral student at the University of California, Berkeley. It showed that for all 2022 U.S. Senate races, U.S. House races and gubernatorial races - where at least 90% of the votes were counted for by Nov. 9 - polls proved to be off by 2.4% on average.
Across the country, the Democratic advantage was estimated at .50 points on average.
"They had almost no overall bias towards the Democrats or the Republicans," Grossmann said. "So what do we know about next time? Nothing. There is no relationship whatsoever between results in the midterm and results in the subsequent presidential election."
From Aug. 18-23, Democratic incumbent Gov. Gretchen Whitmer had the dominating edge in a poll commissioned by the Detroit Free Press and conducted by EPIC-MRA. Fifty percent of respondents selected Whitmer as their likely candidate, while GOP challenger Dixon floated at 39% – and lacked name recognition among non-Republican primary voters.
However, the snapshots of how voters were feeling about the Nov. 8 election started to tell a significantly different story, with:
- A Mitchell Research and Communications poll, conducted on Nov. 3 and commissioned by MIRS, showed 50% of 658 likely voters believing they would vote for Whitmer, 48% for Dixon and 2% still undecided.
- A Nov. 5-7 poll conducted by the Trafalgar Group – with 1,097 recorded respondents – found 48.5% for Dixon, 48.2% for Whitmer and 2% undecided.
- Earlier, American Greatness – a conservative publication aimed at combating alleged Big Media biases – commissioned an Oct. 30 poll from InsiderAdvantage using a 550-person sample. It found 44.8% of participants preferring Dixon, 44.6% backing Whitmer and 5.4% recorded as undecided.
Emerson College Polling, a charter member of the American Association of Public Opinion Research (AAPOR) Transparency Initiative, found from Oct. 12-14, using a sample of 580 "very likely general election voters," Whitmer was selected as the preferred gubernatorial candidate at 49% and Dixon was at 44%.
In an article surrounding the poll, Emerson College Polling Executive Director Spencer Kimball said the poll followed a nationwide trend when it came to statewide polling throughout the country, with "voters who say the economy is their top issue break for Republicans on the ballot and abortion or threats to democracy break for Democrats on the ballot."
Emerson College Polling later conducted a poll after Dixon and Whitmer's final debate during Oct. 26 at Oakland University, collecting from a 900-person sample from Oct. 28-31. Respondents suggested Whitmer holding a 50% lead to the 45% selecting Dixon as their preferred candidate in the race.
"For voters who have heard, seen, or read a lot about the debate, Dixon only trails Whitmer by two points. Whitmer's lead expands to six points among those who have heard a little about the debate, and 13 points among those who did not tune in," Kimball said in the later article.
In the end, however, Whitmer prevailed in the gubernatorial race, defeating Dixon 54.5% to 43.9% – a 468,660-vote difference. After Whitmer's victory, some critics suggested polls inflated Dixon's potential for success over the Democratic incumbent, making the race appear tighter for the sake of having interesting headlines.
On Tuesday, Grossmann said that over the last 20 to 30 years, the average U.S. Senate race has been captured by polls off by six percentage points. Additionally, in the average U.S. House race, polls have traditionally been off by seven percentage points.
"Governor is about the same as Senate – so when you hear a margin of error, you have to at least mentally double that because there's a lot of other factors, a lot of other errors in the survey world," Grossmann said. "The polls were off by 2% on average this year, so that is historically quite good across all of these races."
Grossmann added that he believes the media was reacting to how polls in Michigan have been off for the last two presidential elections, leaning toward Democrats.
"It was not odd for people to think 'well, maybe they might be off again.' They happen to be off this time in favor of the Republicans – slightly – rather than in favor of the Democrats, but it wasn't crazy to think there were sources of error that would continue into the next race," he said.
In 2016, a Nov. 3-4 Public Policy Polling report found former Democratic presidential candidate Hillary Clinton being preferred over past President Donald Trump 46% to 41% with 1,011 likely voters responding. A Nov. 1-3 EPIC-MRA poll additionally found out of 600 likely voters, 42% preferred Clinton and 38% backed Trump.
Trump defeated Clinton in Michigan 47.5% to 47.3%.
"Because of those errors last time, there were Republican pollsters trying to move the numbers for the Republicans – I think partly to influence the media, but partly because they thought they agreed with this story that the polls have been off in favor of the Democrats," Grossmann said. "The things they tried to do didn't work, but I don't know if that should indict polling as a whole."
