Seated next to the window with his steaming cup of coffee and copy of The New York Times, Jim Cunningham — a sculptor and retired MSU faculty member — explains his interest in political polls. He smiles slightly, as he says his desire to understand society is what drives him to pay attention to the polling.
“I want to understand how society is putting forward the people they’re putting forward in this election,” Cunningham said. “The polls are a reflection of the society around me, and I want to understand that society.”
Political polls can be an important source of information regarding voter trends, beyond who is expected to win the election, said Matt Grossman, director of the Institute for Public Policy and Social Research at MSU.
“Surveys are beneficial for learning about the electorate, but we often don’t take full advantage of what they have to tell us,” Grossman said. “They often tell us it’s very close, but they can also tell us what types of people support the candidates and what attitudes led them to do that.”
Coverage of polls, good or bad
The largest percentage of the public says political polling news coverage is good for the election process, according to Pew Research Center surveys.
“I think it’s important for people to know what’s going on and the data behind it,” no-preference freshman Julia Schwartz said.
Most news coverage of political polling reports whom the polls indicate is favored to win the election, Grossman said. This is called the “horse-race” interpretation of polling. The public often complains about this. People say they want more coverage of the candidates’ stances on important issues.
However, Grossman said, studies that offered participants a choice between these two types of coverage found that people are more interested in the surveys.
“People may say they don’t agree with the poll, but they act like they do,” Grossman said.
Founding principal and president of EPIC MRA, a Lansing-based survey research firm, Bernie Porn said the public and the media can become fixated on the horse-race aspect of polling. They miss the deeper understanding polls can provide. Political surveys ask participants questions related to who they will vote for, but also about voting history, their stance on issues and how strongly they hold the opinions they express.
“They stop on the first line,” Porn said. “If you don’t get past the first line, you miss all the important questions about attitudes and why people are voting they way they are. Polling is instructional for the public on what the top issues are.”
Political polls do not typically sway voters, Grossman said.
“Most people have decided on which candidate they are going to vote for,” Grossman said. “They just want to know who’s winning.”
April Kephart, an Ingham County resident, said people either know who they are voting for or they don’t. Some may be convinced if they see one candidate is losing.
However, coverage of the polls can encourage or discourage voter turnout, Kephart said. By seeing one candidate is behind in the polls, supporters of that candidate may be motivated to vote. If one candidate is reported to be ahead, supporters may not turn out to vote, believing it less important.
“If somebody might not have voted but wants Trump, they come out,” Kephart said. “That’s why the Clinton camp is telling people to vote; you can’t just sit back.”
Even though they may pay attention to polls, people are skeptical of polling accuracy, Porn said. The general public doesn’t believe the polls.
“Polls are based on who’s asking the questions,” TJ Szymanski, a regional loss prevention manager for rue21, said. “They’re probably pretty accurate within a percentage. But they’re subject to who’s taking them, the questions, and how the questions are asked.”
Well-conducted polls can be representative of the population as a whole, MSU statistics professor Dr. Pramod K. Pathak said. The laws of probability state that a sample of a larger population can accurately depict the population as a whole, as long as the sample is truly representative.
“Absolutely, they are very accurate— if done properly. They must have validity,” statistics professor, Pathak said. “Validity must be defined as a true cross section of the population.”
However, Pathak said the polls’ findings will not always predict the outcome. Polls predict an outcome, within a margin of error. A typical margin of error is between 3 and 4 percentage points, meaning the actual outcome should fall 3 to 4 percentage points above or below the poll’s prediction. He cautions that though polls are accurate, they may not always prove correct.
“There is no absolute certainty in predictions. There is always a chance they can be wrong. A very small chance,” Pathak said. “Statisticians minimize the probability of the chance of being wrong.”
The public does not generally believe polls are accurate, despite their statistical foundation, Grossman said. However, the public is influenced by them.
“Most people will say that they’re inaccurate, but they pay a lot of attention to them,” Grossman said. “And they tend to agree with them. You ask them who’s going to win the election and they’ll often agree with them.”
Though most may dispute the accuracy of the polls, at least verbally, there are those who believe the polls are good predictors.
“I think they’re accurate to an extent,” Nick Zerilli, a media and information senior, said. “I mean, they’re an actual poll.”
How to use political polls
Political polls can be a useful tool in assessing who is likely to win an election, as well as the attitudes and beliefs that led to that outcome, Grossman said. However, there are some things to keep in mind when looking at the polls.
“You have to see the source, where they come from,” human development and family studies junior at MSU Jewel Dotson said. “That plays a big part in whether the information is credible.”
Pathak also recommends examining the source of the poll. How the poll was conducted directly affects how accurate and trustworthy it is.
“Look at the margin of error first, the sample size and the margin of error,” Pathak said. “And remember there is no absolute certainty. There is always a chance, say a 5 percent chance that polls will be wrong.”
Cunningham said he makes a point to look at multiple polls to get a better understanding of what they are saying.
“They are a rough reflection of societal interest, especially when you pay attention to trends across the polls,” Cunningham said.
This is an important point, Grossman said. Looking at more than one poll helps to provide a better estimate of who is ahead.
“We often cherry pick the polls that support our political views, but that’s really not useful. That’s wishful thinking,” Grossman said. “It’s better to look at the overall average of the polls, what they’re all saying. Don’t just look at an outlier poll.”
The polls can also be affected by the media they were conducted over, Grossman said. Traditionally, polls have been conducted over landlines, because they were in a fixed location. Increasingly, polls are including cellphones. Some are conducted on the internet. This all affects the validity of the sample, which in turn affects the margin of error.
“Learn about how the survey was conducted,” Grossman said. “There’s a big difference in quality and design.”
Polls to pay attention to
For a well-rounded look at which candidate is ahead in the polls, visit http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo.
For a more in-depth look at the local attitudes and opinions behind a poll’s outcome, visit http://www.epicmra.com/pressroom.htm.