William_T._Bianco,_David_T._Canon]_American_Polit

214 Chapter 6Chapter 6 || Public OpinionPublic Opinion

although polls taken early in a presidential campaign (such as a year in advance of the

election) are poor predictors of the ultimate outcome on Election Day, polls taken at

the beginning of the official presidential race, when both parties’ nominees are known,

can provide very good predictions of who will win the election and how many votes he

or she will receive.^36

An alternate technique for measuring public opinion uses focus groups, which are

small groups of people interviewed in a group setting. Because focus groups allow

respondents to answer questions in their own words rather than being restricted to a

few options in a survey question, they can provide deep insights into why people hold

the opinions they do. Candidates sometimes use focus groups to test campaign appeals

or fine-tune their messages. However, because of their small size, focus groups cannot

be used to draw conclusions about public opinion across the entire country.

Large-scale surveys such as the American National Election Study (ANES), which

is conducted every election year (typically both before and after the general election),

use various types of questions to measure citizens’ opinions. In presidential election

years, participants in the ANES are first asked whether they voted for president. If they

say they did, they are asked which candidate they voted for: a major-party candidate

(for example, Hillary Clinton or Donald Trump in 2016), an independent candidate, or

some other candidate.

Another kind of survey question measures people’s preferences using an issue scale.

For a range of topics, two opposing statements are given and respondents are asked to

45 state- and national-

level polls of the presidential race

were released in the last two weeks

of the 2016 campaign.

Source: Pollster.com

DID YOU KNOW?

On average, people should be

more skeptical when they see

numbers. They should be more

willing to play around with the

data themselves.

—Nate Silver, FiveThirtyEight

.com

NUTS

& B O LT S

6.1 The sampling error in a survey (the predicted difference

between the average opinion expressed by survey

respondents and the average opinion in the population,

sometimes called the margin of error) using a random sample

depends on the sample size. Sampling error is large for small

samples of around 200 or fewer but decreases rapidly as

sample size increases.

The graph shows how the sampling error for a random

sample decreases as sample size increases. For example,

in surveys with 1,000 respondents the sampling error is

3 percent, meaning that 95 percent of the time the results

of a 1,000-person survey will fall within the range of

3 percentage points above or below the actual percentage

of the population who hold a particular opinion. If the sample

size was increased to 5,000 people, the sampling error would

decline to 1.4 percent. Mass surveys typically interview around

1,000 people. As the figure shows, this is the point where

adding interviewees provides relatively little improvement

in accuracy.

Sampling errors need to be taken into account when

interpreting what a poll says about public opinion. For

example, suppose a poll of 1,300 Americans found that

66 percent gave an unfavorable rating of Donald Trump’s

performance in office, and 33 percent rated him favorably.

Because the difference between the percentages (33 points)

exceeds the sampling error (about 2.5 points), it is reasonable

to conclude that, at the time the poll was conducted, more

Americans saw Trump unfavorably than favorably.

In contrast, suppose the poll found a narrow 51 to 49

percent split slightly favoring approval. Because the difference

in support is smaller than the sampling error, it would be a

mistake to conclude that the majority of Americans have a

favorable opinion of Trump. Even though more people in the

sample express this opinion, there is a good chance that the

opposite is true in the overall population.

You don’t need to calculate sampling errors to make sense

of political polls—just keep two things in mind. First, large

samples (1,000 or more) are much more likely to provide

accurate information about population opinions than small

ones (fewer than 500). Second, be cautious when you read

about small differences in survey responses, as these patterns

are unlikely to hold true in the entire population.

sampling error

The predicted difference between the

average opinion expressed by survey

respondents and the average opinion

in the population, sometimes called

the margin of error. Increasing the

number of respondents lowers the

sampling error.

2

0

4

6

8

10

12%

100 600 1,100 1,600 2,100 2,600

Sample size

Sampling error

Sampling Error in Mass Surveys

Full_07_APT_64431_ch06_198-231.indd 214 16/11/18 1:38 PM