Texarkana Gazette

The Power of Polls

-

We’ve been hearing a lot about polls as the U.S. presidenti­al election nears. What do polls tell us, and how is the data collected?

The science of statistics involves collecting and analyzing data, or informatio­n, then interpreti­ng and presenting the results. Two common methods of collecting data are

experiment­s and surveys (or polls). There have been and there will continue to be many polls before the elections on Nov. 8. These polls help track the latest trends of Americans’ voting behaviors and candidate choices.

Selection matters

It takes too much time and money to try to survey all 225 million eligible U.S. voters. Instead, statistici­ans (people who gather and analyze data) get a sample (as few as 1,000 or so people) to estimate trends in the population. Scientific opinion polls were started by George Gallup about 80 years ago.

Pollsters care about the quality of the sample, not just the quantity. They will get more accurate results from a random sample that represents the whole population than from a sample taken only from, say, wealthier Americans.

Wording matters

Look at two different versions of the same question and decide which one is trying to make you answer a certain way:

1. “Don’t you agree that pizza tastes better than spinach?”

2. “Which tastes better — pizza or spinach?”

Timing matters

Most surveys are conducted within a window of only a few days. If a survey lasted longer, the last people to respond would have been exposed to news events and candidate remarks that might make them answer differentl­y than if they had been asked at the beginning of the window.

Order matters

Sometimes the order of questions might affect the answers we get. Consider the question “Who do you prefer for president: Donald Trump or Hillary Clinton?” Would your answer be affected if we first asked this question: “Do you think women have equal opportunit­ies to succeed in our country?”

Graphing matters

When you read the results of a poll, make sure they’re not misleading. For example, suppose in a poll Clinton gets 50 votes and Trump gets 45, and the numbers are displayed like this:

The numbers 45 and 50 are actually close. But because the graph begins at 40 instead of 0, when you compare the rectangles, it looks as if Clinton has twice the number of votes!

 ??  ??
 ??  ??

Newspapers in English

Newspapers from United States