The world has changed in terms of how we communicate with one another. Twenty years ago, nearly everyone owned a landline. As cell phones became more affordable and texting became more popular, many people, especially those under the age of 40, stopped using landlines. As of the end of 2015, 48% of Americans had abandoned landlines altogether. Today, older voters, especially retirees, make up the largest swath of reliable landline users. For the purposes of political polling, landlines had traditionally provided the best, most reliable, and accurate voter data. While the data from landlines is still accurate, it’s far less reasonable to get the same results with such a shrinking sample size. Conversely, there are also barriers that pollsters face when targeting cell phones. Most recently, the four main cell providers in the U.S. have begun rolling out a program that will immediately identify political calls, which will allow a larger percentage of the population to ignore these more traditional attempts at reaching voters.
The growing trend now is online polling although there has been some criticism in sampling and varying methodology amongst pollsters. Unlike traditional landlines, nearly everyone of voting age is online. As a society, we are fixed to our screens. Whether we’re watching videos, reading news updates, or firing off an angry tweet, cell phones have become the most important communication tool in the 21st century. If advertisers can target individuals based on cookies and search history, the same principles can be applied to political polling. As we’ve seen over the past few years, online privacy has largely become a relic of the past. We don’t need to look any further than the recent scandal involving Facebook and Cambridge Analytica to understand that this problem is getting worse, not better.
One observation of online polls is that people seem more willing to provide information that they wouldn’t to an interviewer. For example, during the last presidential cycle, Trump fared far better in online polling than polling under the traditional model. This is an example of what is referred to as social desirability bias, similar to what we more commonly refer to as the Bradley effect. In this scenario, a respondent reports an answer in a way they deem to be more socially acceptable than would be their “true” answer. Another example of this was Brexit, where traditional polls favored Britain remaining in the EU, while online polling favored leaving. Due to the assumption that traditional polling is more reliable and accurate, there was little consideration that online polling would correctly predict the results for both Trump and Brexit.
Pew Research did an interesting study about the subject of social desirability that demonstrates the difference of what one is willing to admit in a non-judgmental Internet poll versus a live interviewer. Knowing that people are less likely to be honest with live interviewers in conjunction with the shrinking universe of landlines, it is important that moving forward, online polling be given greater credence than they have in the recent past as long as the sampling and methodology are up to muster. With any poll, the results can be interpreted to fit the original hypothesis, so if you don’t see the cross tabs, hold off on accepting them as gospel.
At Grassroots Midwest, we find a lot of value in 5 question robocall polls where ALL likely voters are contacted. Depending on the district, those can yield at least 1,000 responses that is typically proportional by geography. These polls tend to skew a bit older because we are targeting landlines, but because our targets aren’t interacting with live questions, we bypass any problems that could arise from social desirability bias. In many cases, these turn out to be more accurate than some of the scientific polls that are popular. It was through these simple polls that we saw the path for Trump to win Michigan despite what the experts predicted.
The more you know about your electorate, the more effective your campaign will be. If you or your organization are interested in any variety of polling or voter targeting, we can provide you with a number of options to meet your needs.