During political elections, news organizations often use public opinion polls to help gauge which candidate is the front runner, and why. University of Michigan's Dr. Vincent Hutchings explains the science of random sampling that makes it possible to query a few hundred or thousand people and use that data to accurately determine how the general public might vote. "Science Behind the News" is produced in partnership with the National Science Foundation.
Science Behind the News - Opinion Polls & Random Sampling
ANNE THOMPSON, reporting:
The 2012 presidential campaign is in full swing, with Republican candidates battling state-by-state to determine who will face President Barack Obama on Election Day.
BRIAN WILLIAMS (File Footage): In the poll numbers we are debuting tonight, there is a new GOP frontrunner in this race.
THOMPSON: A crucial part of this grueling electoral process is polls, the almost daily snapshots of public opinion that help measure who's up and who's down among the candidates.
Dr. VINCENT HUTCHINGS (University of Michigan): The aim of an opinion poll is to get a sense about a population and where they stand on some particular set of issues or policies to measure their attitudes in effect.
THOMPSON: Vincent Hutchings, an NSF-funded professor of political science at the University of Michigan, says that public opinion polls rely on a concept in statistics known as “random sampling,” the idea that it's possible to draw a clear picture about the feelings of a large group of people by examining how a small, randomly-assembled slice of that group feels.
HUTCHINGS: The process involves interviewing a very small subset of that population and doing so in a scientific manner so that you have a better accurate sense, but more efficient sense of what the entire population understands.
THOMPSON: An opinion poll might focus on a particular state such as Florida or the entire U.S. population of more than 300 million people. Pollsters use computer programs to generate a random list of a few hundred or a few thousand telephone numbers from the larger group and then they call each number to survey people for their opinions. Hutchings says making a random sample is a lot like cooking soup. You don't need to eat the whole pot to know if it tastes good, you just need a spoonful.
HUTCHINGS: The cook merely needs to get a spoon out, you know, taste it. That spoon represents a sample, as it were, of the actual contents of the pot.
THOMPSON: Samples in general give better poll results when they include more characteristics of the whole population, and in the same proportions as the population. Back to the soup analogy, if you're cooking chicken soup, you want your sample to include the ingredients, such as chicken, noodles and broth, in the same proportions as the overall soup.
HUTCHINGS: We want to make sure that the sample has the population characteristics that are demonstrated in the larger population to which we want to make an inference.
THOMPSON: When reading a poll, it's important to also study the fine print, usually at the bottom. You'll find information there about how the poll was conducted and what the size of the random sample was. You'll also find the margin of error, a number with a plus or minus sign in front of it. This number tells you the range of accuracy of the poll, in this case three to five percent points. Typically, the smaller the margin of error, the more accurate the poll is.
HUTCHINGS: What that means is that the number that's reported in that survey, we have a sense that given the size of the sample and given the level of uncertainty associated with that size, the number could be three to five percentage points higher or lower. But we know it's in that range.
THOMPSON: Even though the polls rely on just a slice of the population to gauge public opinion, they are far more accurate than you might think, which is one reason why they play such a special role in politics.
HUTCHINGS: Public opinion polls provide us with a way, absent an actual election, to discern where the public stands on various issues.
THOMPSON: As Election Day 2012 draws closer, the science of public opinion polls will help give a clear snapshot of who might be our next president.
Polls have long been the gold standard for assessing politicians, elections and voter concerns. They haven’t been shining lately. Their predictions were far off the mark in both the Scottish independence referendum and the 2014 U.S. congressional elections.
Public, Opinion, Polls, Polling, Random, Sample, Sampling, Political, Candidates, Campaigns, Elections, Voters, Voting, Vincent Hutchings, University of Michigan, Population, Survey, Data, Information, Numbers, Statistics, Stats, Percent, Percentages, Pollster, Issues, Ballot, Characteristics, Democracy, Range, Margin, Error, Margin of Error, Average, Predict, Prediction