We all have a love/hate relationship with polls. The “Pollercoaster” can be exciting when it gives new numbers on the state of the race, but it also can be a wild ride given its unpredictable nature. Polls are one data point and can be important in understanding what’s happening if used correctly. Unfortunately, they are often misinterpreted, or worse, purposely misconstrued to push a narrative. Here are some tips about how to ride the pollercoaster and how they can be used in 2024.

Not All Polls Are Created Equally

In recent years, there have been an explosion of new polling outfits that claim to be doing legitimate polls. Some of these are no doubt valid polls that are worth considering. At the same time, there are some very questionable polls that have put results without explaining their methodology and are clearly designed to push a narrative.

Helpfully, 538 has updated its pollster ratings to reflect past performance, the amount of disclosures about methodology, among other metrics. Ultimately, not all polls are created equally and some are going to be inherently better than others. It’s also important to keep in mind why the poll was commissioned, is this a non-partisan poll released by a news organization or is this an internal released by a candidate or advocacy organization.

These are very basic things that long-time observers of politics know, but it is still important to keep in mind when you see a poll. It is all too easy to jump to conclusions without taking a moment to consider where the poll is coming from. Ultimately, the old adage sometimes is the best, if a poll is too good to be true, then it probably is.

That’s why it can often be helpful to look at the average of polls , which is created by created by combining the results of recent polls. That way one outlier poll is contextualized and is not the only thing driving the conversation. While polling averages also have their limits, it is another way to look at the totality of the polling landscape, instead of individual polls.

Cross Tabs Have Their Limits

Perhaps no topic is more controversial than cross tabs. These are subgroups within polls that break down the poll results based on groups such as different races and ages. Cross tabs will be dissected by all sides to either make a point about a particular subgroup or to prove the poll is not valid because of cross tabs results.

The most important thing to remember about cross tabs is sample size. This is the number of people who participated in a poll. Generally, the larger the sample size, the more accurate the poll result will be. So a poll with 1,000 people is better than a poll with 500 people. This is particularly true with the cross tabs and care has to be taken not to extrapolate too much from a such a sample size, because remember cross tabs are only a small portion of the overall poll. In other words, a poll itself is a sample of the overall population, and a cross tab subgroup is an ever smaller sample of a particular group.

This often happens when you see cross tabs of polls showing a surprising result such as Trump winning 18-29 voters. The problem is the sample size is likely very small. For example, a poll might have 1,000 total respondents. One subgroup, such as 18-29 year olds, might be 10% of all respondents. So the sample size of this subgroup is only 100 people. This is almost certainly too small to make any meaningful conclusion, and usually at least 500 people are needed for a minimum sample size.

That’s not to say there is no point to cross tabs or you can never look at them because they are one way to determine what’s going on in the poll. However, it is hard to make a direct connection between a single cross tab subgroup and generalization about that group. In order to understand the opinion of particular subgroups, looks for polls with a large sample size that occasionally come out that are only polls of a particular subgroup, such as Hispanics.

Uncertainty and the 50% Rule

Polls are often presented as Candidate A leading Candidate B by 5 when the polling result is 45%-40%. While that is true, that means 15% of voters are unaccounted for. Maybe they are undecided, maybe they won’t vote, maybe they just didn’t want to tell pollsters who they will vote for. All that creates uncertainty in the outcome because that 15% can swing the election. Generally, the closer a candidate gets to 50%+ in a poll, the more confidence there can be that candidate is in a position to win. After all, even if all the undecideds go to the other candidate, the candidate at 50% will still win. That’s not to say the poll is wrong or voters won’t switch their mind, but it gives a bit more certainty.

At the same time, it doesn’t make sense to dismiss any poll with a leading candidate below 50%, and in fact, most close races will not have a candidate reaching that threshold. However, there are been an increasing number of polls with over 15% of voters are not supporting one of the two main candidates. These types of polls might look something like Candidate A 40%, Candidate B 37%, and other 23%, even in a match up of Biden vs. Trump. It is hard to know how useful a poll like that will be because it’s almost certain that Biden+Trump will get at least 95% of the actual vote. That doesn’t mean totally toss out the poll, but it also doesn’t mean that poll has to be viewed as super helpful.

The Pollercoaster in 2024

Applying these rules to 2024, we can come away with a few conclusions about polling and the presidential election. There have been a number of high quality polls showing Trump with a lead, and so Biden’s struggles cannot be dismissed because they are pretty consistent. That doesn’t mean that Trump will automatically win, but it is does mean that Biden is facing a challenging environment. 

At the same time, the use of the polls have not always been done in a way to convey their nuance and limitations. For example, cross tabs that are showing Biden losing young voters or Hispanic voters have to be taken with a grain of salt. Even if Biden is struggling, it’s unlikely that he would be bleeding so much support that would lead to these results. The polls shouldn’t be over interpreted or be used to show something that is not really supported by the results.

The biggest issue is that neither Biden nor Trump is usually getting close to 50% either nationally or in the most important battlegrounds. Trump is usually closer to his 2020 numbers, but rarely goes above them and seems to have a ceiling with them. Biden is further back from his 2020 numbers and there are usually a good amount of voters who are uncertain about what they are going to do. This unpredictability is not necessarily good news for Biden, it also gives him the hope that he can win them back as the election gets closer

For example, when undecideds are forced to pick (pushed) , this often helps Biden because he likely has more soft supporters at this point. This may be because the thinking goes that Biden does better with “double-haters,” a group of voters who don’t like either candidate. In addition, a decent chunk of voters currently claim they will vote for a third-party candidate, but that almost never ends up happening in a Presidential election. Finally, it is likely that a number of voters will make their decision at the last minute or perhaps just stay home.

In such an environment, it seems too early to make any final conclusions about what’s going to happen and a lot can still change. The polls are a snap shot in time and polls are usually a lagging indicator. 2024 is going to be a wild ride and we all need to be prepared to see where it takes us. As the election gets closer, the polls likely are going to be more accurate as the number of undecided voters tend to decrease.

So by all means, take a look at the polls and use them as a data point about the current state of the race. But it is important not to get so obsessed with the minutia of polling data that the bigger picture of what’s happening the election is lost. Ultimately, the election will be decided when people go vote, and not just by the polls.

*Sam Shirazi is an Attorney who lives in Arlington, and is a Virginia elections analyst.