From the course: Everyday Statistics, with Eddie Davila

Political polls

- [Instructor] In politics, poll results are currency. They're used to raise money, motivate candidates, identify and sway undecided voters, and also, to tell a story. More than ever, statistics drives politics, which often makes people wonder, with so much money and effort behind these polls, why does every poll tell a different story? Political polls are based on surveys of people. People might be approached by a pollster on the street. They might be called at home or on their cell phone. In some cases, people may complete an online survey. Each polling group develops their own method for gathering data. It doesn't take a lot of statistical know-how to realize data collection is sometimes flawed. Lots of people can easily escape being polled, and some are less likely to answer certain types of surveys, so some groups may be very underrepresented in certain types of polls. And even those that are polled might still be undecided or some people may be for one candidate today but still considering another. Others know exactly who they'll vote for, but they're too shy or embarrassed to tell the pollster. And of course, who's to say that any of these people will actually show up to vote? With so much possible variability, it's important to consider each polling organization and their methods. Also consider the influence of polls in raising money, swing voters, and even in identifying opportunities. Thus, there are many different polling organizations with different agendas. There are also some different groups that value different polls at different times. Some groups use polls to signal a need or opportunity to raise money, others use polls to signal strength to undecided voters. And even a leading candidate may wanna publicize a poll that makes them look vulnerable in an effort to motivate their base to actually show up and vote. By changing their methodology, a poll could purposefully or just accidentally overrepresent one demographic and completely miss others. And don't forget to consider the questions asked in these surveys. The phrasing of the question and the available answers to each question might provide very different outcomes from the very same group of people. Nonetheless, a good statistician can look through dozens of different polls, understand the individual methodologies and the questions used, and sometimes, together, all of these polls might tell an interesting story. An interesting story for statisticians, anyway. Next time you see polling results, slow down and consider what's being reported, and not just what's being reported, but how the data was collected. Use your statistics foundation to help you better understand the different pieces of the political puzzle. If you wanna know more, check out my Statistics Foundations course in the library.

Contents