In Virginia, every year is an election year; this means you’re likely to see polls year-round. Many people are unaware of how polling works and how much effort goes into it.
10 News spoke to Dr. Harry Wilson, the senior political analyst for the Institute for Policy and Opinion Research and professor emeritus of political science at Roanoke College, about polling and the most common misconceptions the average reader has.
Recommended Videos
10 News asked Dr. Wilson to give us the most simplified steps behind the average polling cycle, which are as follows:
- Writing the questions
- Sampling & data collection
- Data analysis
- Writing the press release
Typically, Roanoke College collects data (usually over the phone and text-to-web) over the course of four to ten days, but this can vary. Staffing, how many interviews are set to be completed and scheduling conflicts can inflate the amount of time it takes to conduct a poll.
Dr. Wilson tells us that the data analysis can be completed in as little as 24 hours, which can be crunched further if it is a time-sensitive poll, such as the final one before an election.
Roanoke College’s polling system also utilizes quotas that ensure regions are accurately represented, so areas with higher populations have more respondents than areas with lower populations. After the data is collected, they will statistically weight the poll by age, race, gender and sometimes political party.
Despite the amount of work and research that goes into a poll, they’re rarely 100% accurate, and pollsters can have trouble nailing expected turnout. Most recently in 2025, Zohran Mamdani pulled an upset victory in the New York City Democratic mayoral primary against former governor Andrew Cuomo, when most polls predicted an easy victory for Cuomo.
According to Politico, this can be attributed to the massive influx of voters who haven’t recently voted in a primary. Dr. Wilson told us that there is no sure-fire way to predict turnout for an election.
“If anyone has a crystal ball to accurately predict turnout, please lend it to me! Again, there are different ways of doing this. You can ask how likely someone is to vote and assign different weights to different responses. You can also include a calculation based on their past voting record (do they only vote in presidential elections, statewide elections, etc.). We use the most recent applicable exit poll data. So, for the November Virginia gubernatorial election, we will use the exit poll data from the 2021 election. That is, at least, an objective standard.”
Dr. Harry Wilson, senior political analyst for IPOR and professor emeritus of political science at Roanoke College
Roanoke College has also had its missteps in polling, such as in Roanoke’s 2024 mayoral race, when it predicted an easy victory for Democratic candidate Joe Cobb. By the final vote, according to the Virginia Department of Elections, only 59 votes separated him and his Republican rival, David Bowers. Dr. Wilson said he doesn’t have a perfect explanation for why the polls were so off, but he suggested that polling in lower turnout elections is oftentimes more difficult.
“What was very strange about that poll was that we were pretty darn accurate on the council races and so far off on the mayoral contest. I don’t have a good explanation for that, but I sure wish I did. Polling at the local level is trickier, and we obviously did not get that one right. Those are very low turnout elections, so reaching people who will actually vote is very difficult.”
Dr. Harry Wilson, senior political analyst for IPOR and professor emeritus of political science at Roanoke College
While Dr. Wilson acknowledged that pollsters can make mistakes, he mentioned the importance of ensuring that a poll is as accurate as possible, as an incorrect poll can damage the reputation of a pollster.
“Some people seem to think that we make this stuff up or we manipulate the results to get an outcome we desire. The outcome I want to see, regardless of the poll or my personal preferences, is the most accurate one. The reputation of Roanoke College and the RC Poll is on the line, to say nothing about my personal reputation and integrity. All of those matter a great deal to me.”
Dr. Harry Wilson, senior political analyst for IPOR and professor emeritus of political science at Roanoke College
There is also always a margin of error for any poll conducted, which can be based on the number of completed interviews, the size of the population and statistical weighting.
For the college’s final poll in the 2024 general election, the margin of error was 4.6%, with a confidence level of 95%, which means around 95 out of 100 samples should be within 4.6% of the predicted value. The final poll showed Kamala Harris with 51% and Donald Trump with 40% of the vote in Virginia, while the actual election results showed Harris at 51.8% and Trump at just over 46%.
It is also worth noting that many political polls are conducted statewide as opposed to nationwide, and swing states, such as Pennsylvania or Wisconsin, usually have more polls conducted than safely Republican or Democratic voting states.
Dr. Wilson also mentioned the belief that many people hold, which is that polls are falsified, biased or a sort of propaganda. He tells us that it is the job of the pollster to be as accurate as possible, even when it is being conducted by a news source many see as having a political slant, such as CNN or Fox News.
“People also assume that a poll conducted by Fox News will reflect that network’s bias as will a poll done by CNN. In fact, they are both trying to ‘get it right.’ The polling groups are separate from the news/editorial groups.”
Dr. Harry Wilson, senior political analyst for IPOR and professor emeritus of political science at Roanoke College
Much like a reputable news station, the job of a pollster is not to project their bias onto the public, but rather to show how the population might vote in a given election. Dr. Wilson encouraged poll doubters to look at the track record of any given pollster and see their accuracy before dismissing their efforts.
“Look at the track record of the polling firm. I’m a big fan of the Pew Center. That doesn’t mean that I always like (in the sense that I agree with) their results, but it means I trust they did things correctly and objectively...”
“I also know the director of polling at Siena College, who does polling for the New York Times. They do things the right way, and they care a great deal about their accuracy.”
“Are there bad polls? Of course, but there are a lot more good ones. None of us are perfect, but the vast majority of us are doing the best we can. We’re trying to predict human behavior from a sample of people. Even if we do everything correctly, we can be ‘wrong’...”
Dr. Harry Wilson, senior political analyst for IPOR and professor emeritus of political science at Roanoke College
Pollsters aren’t always spot on, but most attempt to be as close as possible to give an accurate depiction of what an electorate might look like in a general election, or how the public feels about a certain candidate. Pollsters may also ask for the favorability of individuals who are currently in office, what a voter’s top issues are and if a respondent has voted previously. While many polls can be a few percentage points off, the only ones that can elect someone into office are conducted in the voting booth.