How To Read A Poll

Key Takeaways 

• Public opinion research can offer leaders useful insights. 

• But polls should be interpreted narrowly and carefully. 

• Polls can overestimate how much the public knows and understands your topic. 

Political and commercial organizations often rely on polls. The promise of survey research is that it will allow leaders to know what people think about a wide range of topics. Companies can use the research to offer more appealing products, or to respond to employee needs and concerns, making their organizations more efficient. Media outlets can use surveys to define the boundaries of public discourse. And, of course, politicians could use surveys to better represent their constituents. 

Unfortunately, it is not that simple. Surveys can offer useful insights, provided that they ask appropriate questions that capture real, durable opinions and those results are interpreted in a narrow, limited fashion. Leaders can use surveys as a guide but should be careful and think about the limits of public opinion. 

Even the best survey methodology must confront the problems inherent in studying human beings. Almost 30 years ago John Zaller’s research on public opinion revealed that most people do not have opinions about most issues. He found that many survey responses are just “top of the head” answers. Most survey respondents guess at the answer based on context clues in the question and whatever happens to be most easily accessible in memory at the time, much like an undergraduate guessing his way through a multiple-choice test. 

The problem is that the world is very large and human beings have cognitive and attention limits. To have an opinion about something, I must care enough about it to pay attention, learn something, remember that thing, and form a view of it. Ask me about something that I don’t care about, and I may give you an answer, but the answer does not mean anything. The “opinion” registered in the survey only exists because you asked, and I felt compelled to give you an answer. Ask me the same question a week later and my answer could be completely different because I do not really have an opinion about the issue.   

We can see how the public’s lack of understanding can lead to misleading poll results when we examine differences in question wording. For example, recent polls show that Americans support imposing a no fly zone in order to aid Ukraine in its struggle against the Russian invasion. However, that support drop considerably when respondents were informed that doing so might involve shooting down Russian planes and is considered an act of war.   

In other words, Americans are broadly supportive of Ukraine, but have a poor understanding of international relations and the effects of various military actions and economic sanctions. It is useful to know that Americans support Ukraine. It is foolish to assume that Americans have a firm understanding of the risks and costs of various actions.  

In general, the more well-known and important something is, the better the chance that people have a “real” opinion of it, but even these opinions should be viewed in a larger context. For example, President Biden has near-universal name recognition and people recognize that the presidency is important. Most Americans can offer an opinion on the president. 

However, that opinion may be offered without any knowledge the president’s job performance. Partisanship, of course, is a key heuristic for most voters, and survey results should be interpreted with that in mind. You can see this partisan bias in the wide disparity between Republican and Democratic views of the president. Because the president is such a large figure, and he is viewed as an avatar of his political party, to a large extent presidential approval polls don’t measure opinion of the president; they just measure partisanship. 

So how should you read a poll?  

Carefully. 

When you read the results of the poll, do not just take it at face value. Consider how the respondent viewed the topic. Do people know anything about this? Do they care about it? Is there something that might have biased their responses? To make use of survey data, think about how the data are generated and the larger context of that information. Weigh the information according to how much you think that respondents gave you real, clear, stable opinions that you use to inform organizational decisions.