How can polls work if people won’t respond?
Q: How do polls work if people refuse to answer them?
A: I have written in the past about how polls are simply a snapshot of public opinion at a particular moment, not a projection or prediction. But, here I would like to discuss the issue of who answers a poll, or sampling. The idea of a poll is to ask a small number of people a series of questions to get a better understanding of what a larger group of people believe.
In a poll we ask a small group of people, called the sample, in hope of understanding a larger group, which we call the population. The underlying premise of polling is that the sample can scientifically reflect the population (within a margin of error) when it is properly selected. When I teach polling, I often use the analogy of a bowl of soup. A person can tell the flavor of the soup with just a few spoonfuls, as long as all of the ingredients that are in the soup make it onto the spoon.
The way we do this is through probability sampling. It basically means that I can get a representative sample if I give each person in the population I’m studying an equal chance of being selected. The traditional way to do this is to randomly dial phone numbers of people in the population you are surveying, and since most people have phones the results are largely a random sample.
What happens when people steadfastly refuse to participate? Problems. Non-response bias is a formidable opponent in polling, especially when important groups of voters collectively refuse. However, it is not new. There have always been individuals who, for various reasons, opted not to participate. Whether driven by a distrust of institutions, a desire for privacy, or sheer disinterest, these non-respondents can hurt the accuracy of polls.
Pollsters have developed a set of tools to mitigate the impact of non-response on the overall reliability of their findings. One such tool is the art of weighting. By adjusting the results to reflect the demographic characteristics of the population, pollsters can compensate for the imbalances introduced by non-response. This method involves assigning greater importance to the responses of certain groups, ensuring that the final results accurately mirror the choices of the voters most likely to turn out.
Technology has helped. Online surveys, automated calls, and interactive voice response systems offer alternative avenues for collecting data, reducing the reliance on traditional phone-based polling methods. This diversification of approaches allows pollsters to cast a wider net and capture a more representative sample of the population, even in the face of non-response. Researchers use lessons from psychology, sociology and communication studies to better understand the motivations behind non-participation and tailor their approaches.
However, it is still a challenge. The rise of caller ID, spam filters, and a public that is increasingly reluctant to engage with strangers all contribute to the growing pool of non-respondents. Getting that representative sample can be difficult and more expensive.
Kevin Wagner is a noted constitutional scholar and political science professor at Florida Atlantic University. The answers provided do not necessarily represent the views of the university. If you have a question about how American government and politics work, email him at kwagne15@fau.edu.