Measuring public opinion 217make you suspicious that pollsters are trying to skew their findings—rather, they show
just how hard it is to accurately measure opinions. (See Take a Stand for more on public
opinion on immigration.)Unreliable Respondents Another problem with surveys is that people are
sometimes reluctant to reveal their opinions. Rather than speaking truthfully, they
often give socially acceptable answers or answers that they believe the interviewers
want to hear. In the case of voter turnout in elections, up to one-fourth of respondents
who say they voted when surveyed actually did not vote at all. Political scientists
refer to this behavior as the social desirability bias, meaning that people answering
survey questions are less willing to admit to actions or express opinions, such as racial
prejudice, that they believe their neighbors or society at large would disapprove of.^37
Pollsters use various techniques to address this problem. One approach is to ask
questions in multiple ways; another is to verify answers whenever possible, such as
checking with county boards of elections to see if respondents who said they voted
actually went to the polls. When there is concern that respondents will try to hide
their prejudices, pollsters sometimes frame a question in terms of the entire country
rather than the respondent’s own beliefs. For example, during the 2008 and 2016
presidential primaries, rather than asking respondents whether they were willing to
vote for a woman candidate (such as Democrat Hillary Clinton), some pollsters posed
the question indirectly, asking whether a respondent believed that the country was
ready for a woman president. Another tactic is to ask respondents for a different kind
of evaluation—rather than asking whether they like or dislike candidates, pollsters ask
whether they would sit down for a beer (or coffee) with them.
Of course, attempts to account for the social desirability bias by rephrasing a
question can introduce new biases. For example, a survey question designed to ask
about a sensitive topic might begin with “Some people have argued that.. .” However,
in some cases adding this phrase to a question can produce higher levels of respondent
agreement regardless of what is being asked. Again, the problem is not that pollsters
are dishonest or inept—the problem is that public opinion is often difficult to measure.The Accuracy of Public Opinion Because poll results play such an important role in
media coverage and public discussions about politics, questions are often raised about the
assumptions and corrections that pollsters use to account for the inevitable ambiguities
related to poll results. During the 2016 presidential campaign, many Republican politicians
and observers argued that polls showing a narrow but significant lead for Democrat Hillary
Clinton over Republican Donald Trump were the product of inaccurate assumptions about
turnout. Some even argued that pollsters were deliberately skewing their findings so as to
bolster support for Clinton. Of course, Trump won the election, and it appears that pollsters
did misestimate turnout of some groups. However, this outcome does not imply that
pollsters altered their surveys to hurt Trump’s chances. The same models that mispredicted
turnout in 2016 have proved reliable in many other elections; that’s why pollsters used
them in 2016. Their mistake highlights the difficulty of accurately surveying the American
people and shows why we should be reluctant to overinterpret survey results.
In some cases, inaccurate or outlandish survey results are generated because some
respondents don’t take surveys seriously. They agree to participate but are not interested
in explaining their beliefs to a stranger. Faced with a long list of questions, they give quick,
thoughtless responses so as to end the interview as quickly as possible. Misperceptions
may also result when respondents form opinions on the basis of whatever considerations
come to mind. And most importantly, people may reach for considerations that are easy
to use but not very informative. Consider claims about Obamacare. A 2015 poll asked
people whether health care spending by the federal government was higher or lower thanFull_07_APT_64431_ch06_198-231.indd 217 16/11/18 1:38 PM