What the polls can tell us

October 28, 2024

Two long-term politics watchers weigh in on why exit polling and winner projections are an essential part of the political process, especially in challenging election years.

Graphic about political polling
Graphic by Violet Dashi

Starting with the 1968 elections, a young Michael Montgomery spent election nights with his parents in the WJR studios in the Fisher Building. His dad, Al Montgomery, had a side business calling election results through a statistical model they all worked on. For decades, they projected winners at the state-level, like Governor Bill Milliken over Sander Levin in 1972, and the national one, like President Jimmy Carter over President Gerald Ford in 1976.

Montgomery, now a 蹤獲扦-Dearborn Health and Human Services lecturer, and his parents made their predictions in the same studio that Hermits Cave a Detroit radio show with eerie sound effects and spooky storytellers that became nationally syndicated in the 1940s was once broadcast. As a kid around age 10, I thought it was marvelous. There was a little door that was used to make the radio shows squealing sounds right in that studio, Montgomery says. 

Almost 60 years since that 1968 election night, there is a different type of spooky on the horizon. And it has nothing to do with Halloween. With election day less than a week away, its the distrust in the election process. A recent  by the polling firm Gallup found 19% of participants had no trust in the legitimacy of the election result. In 2004, this figure was 6%.

Its been a few decades since Montgomery has been involved in calling a political race and he says hes seen divisions widen significantly in that time due to party differences with issues like civil rights, womens rights and health care access. The electorate became even more divided four years ago, say both Montgomery and Communication Professor Tim Kiska, whos called elections for media outlets for 40 years.

Sixty years ago, we made these predictive models because news outlets wanted to be the first to give people an update on who won. It was exciting to be in that studio for WJR and be able to report breaking news that was so important that it interrupted a ball game. We didnt know it, but we were helping an expectation that winners and losers would be known on election night or in the early hours of the morning, says Montgomery, whose parents transitioned out of the polling business in the 1980s. In too-close-to-call elections, it would be better to wait until all ballots were counted, which could take a couple days. However, this expectation of a quick turnaround has been set. If the news is delayed beyond expectation, there is now a likelihood that wild conspiracy theories will take off. So its imperative to be fast and accurate.

Among concerns of distrust and political polarization, the two professors sat down to discuss what goes into election predictions and calls, along with the importance of polling and why it matters.

Exit polling

Weve all seen them the people who stand 10 feet from the building where we vote and ask for our time to fill out an anonymous survey. But who are they and should we take the time to stop?

Kiska says exit polls are conducted by companies like Edison Research, which does polling on elections across the globe, and that what people say in them today gives insight into election results now and far into the future. We can talk until we're blue in the face, but what issues were important to voters in a given year?. Abortion? Immigration? We can speculate, but exit polls tell us directly from the voter through an exit poll survey to fill out on what issues guided their vote, Kiska says. 

Kiska says if a voter has time and is asked to fill out an exit poll while leaving their precinct, they should strongly consider taking it. 

The information gained from exit polling is political scientist gold. Its a record of what groups voted for whom like, did more women vote for Harris or did an older demographic vote for Trump? that can give insights on election night and for 100 years down the road when we ask ourselves, What the hell happened here?, Kiska says. These exit polls give statistics that create a great historical dataset of what happened in an election thats directly from the people theres nothing better than that.

Precinct polling and forecasting

According to FiveThirtyEight an ABC News-affiliated website that uses polling as well as economic and demographic data to explore likely election outcomes the 2024 presidential candidates are tied. But on election night, once the ballots start coming in, predictions can begin to be made by looking at voter trends and the demographic makeup of a precinct. Once adding in other key election precincts, a pattern often emerges to see whos ahead. Thats precinct polling.

Montgomery did this work from the late 1960s to the 1980s; Kiska began precinct polling in the 1970s while at the Detroit Free Press his first presidential election was Gerald Ford vs. Jimmy Carter.

Kiska says it is often done like this: A news outlet sends a representative up to a multitude of voting precincts at 8 p.m., when the polls close, and waits for results to come in. Once they do, the person calls in the voting results from the specific precinct, often selected to get a diverse set of demographics that accurately represents an area. Lets say we get numbers from 80 selected precincts. If you put them together, youll get a pretty good indicator on whats happening, he says. New state laws allow absentee ballots to be processed up to eight days before Election Day and counting to begin as early as 7 a.m. on Election Day (although results cannot be released until 8, when the polls close) will also help with making an accurate prediction, Kiska adds. He explains that those numbers later combined with general election results will then be analyzed by mathematicians and political scientists using a variety of models to get predictive percentages.

Kiska says the precinct polling principles used today are very similar to  when he began this work in the 1970s. But there is one big difference: The computers may have made all the difference in the world, Im sure the firepower we have in most laptops is greater than all of NBC News in 1976.

The big question: When it comes to election announcements, is polling trustworthy?

Both professors say yes.

Kiska says no one wants to be wrong. First, as Montgomery also mentioned, errors fuel conspiracy theories. There's an X factor here that didnt exist decades ago and wasn't as obvious four years ago people are looking for evidence of shenanigans, even where there isnt any, says Kiska. We need all the tools we have and to be accurate and efficient with them to avoid the speculation that theres wrongdoing.

Second, if an error is made, it can ruin someones reputation. Kiska speaks from experience. In my 40 years calling elections, I  but recalled it a couple hours later. The only reason my reputatio