Public Opinion


Everything in politics today is dominated by “polls”. We are going to spend half of this class talking about polls. That may seem like too much, but actually we could spend an entire course on this topic. It is that important.

Elections are now often decided long before Election Day. Candidates decide whether to run for office based on polls months, even years, before any real votes are cast. Politicians follow poll numbers the way people used to follow the stars through astrology.

The petition drive to recall California Governor Gray Davis was based on polls showing his low approval rating. The approval rating is the simplest, and most reliable, indication of a politician’s fortunes. “Do you approve or disapprove of the job so-and-so is doing?” There is little bias in the question, and everyone seems to understand it. The lower the approval rating, the more vulnerable the politician is. Gray Davis’ approval ratings were in the 20-30% range. That’s about as low a rating as anyone ever sees.

President Clinton’s approval ratings dipped into the 30s (e.g., 38%) in his first year in office. His health care plan failed, and he had the disaster at Waco. Republican candidates for Congress in 1994 began to run against Democratic incumbents by linking them to President Clinton. Because his approval rating was so low, Republicans won a landslide and took control of Congress in 1994.

After the terrorist bomb destroyed the federal building in Oklahoma City, Clinton’s approval ratings began to climb. By 1996, he had strong ratings again and easily won reelection.

When Clinton was impeached by the House of Representatives in 1998, many expected his approval rating to fall again. But it didn’t. The public continued to support him, and enabled him to remain in office. In many ways, the poll numbers were as important as the opinions of congressmen.

Should poll numbers be so influential? Can they be misleading? Can they be manipulated? We’ll discuss these issues. Then we’ll take a look at what the public thinks about certain issues.

What is a Poll?

A poll is simply a measure of the view of a group of people. I could take a poll of the opinion of this class. That would be cheap and quick. Or I could take a poll of what people in my town think about something. That would take more time and money.

Would I have to ask everyone single person in my town in order to estimate its views? Finding and talking to every single person would improve the accuracy of my poll. But even that would not be 100% accurate, become people might change their mind or fail to tell me, a stranger, what they really think.

If I only need an estimate, and can live with the possible of a small error, then I do not have to ask every single person. On a political issue, I could obtain a good picture of my town’s view by asking a random group of 40 Democrats, 40 Republicans and 40 Independents. That is only 120 people in all. I could call them over a three day period.

That is all that most political pollsters do. For nationwide polls, the group running the poll will only call a thousand or so persons. This is called “random sampling.” The sample is carefully selected, however, to make sure it represents the nation as a whole. For example, there should be the same percentage of union workers in the sample as in the nation at large. There should be the same percentage of religiously devout people in the same as in the nation.

In predicting elections, the most difficult part is finding a sample that represents people who will actually vote on Election Day. Half or more of eligible Americans do not typically vote. If you call someone at random, the odds are 50/50 or worse that you are talking to someone who does not vote. His or her opinion is irrelevant. It should not be included in a political poll. The lower the turnout – e.g., the higher the percentage of those who do not vote – the more difficult it is for a poll to predict the outcome of an election.

Query: which can polls predict better, general elections or primaries? (Answer: general elections, because they have higher turnout.)

In “straw polls” that invite the public to contact the pollster, as on the internet and the American Idol, there is no control over the sample. The people who take the initiative to give their opinion to the pollster are not representative of those who do not. Also, multiple voting by the same person can occur. These are unreliable polls.

Polls are constantly improving how they select their samples, but can never choose a perfect sample. If you are interested in math, then you may want to look for books and articles on this topic.

How Polls are Manipulated

Poll results are easily manipulated by distorting how the questions are asked. In a poll on gun control, the question “Do you support gun control?” will obtain more “yes” answers than the corresponding “no” answers to “Do you have a right to own a gun?” So gun control proponents will ask the first question in their polls, while gun rights proponents will ask the second question in their polls. Both will claim that the public supports their view.

Ordering of questions will also change the answers. Ask the question “Do you think convicts should own guns?” before asking a question about gun control, and you will increase a response in favor of gun control. In contrast, ask the question “Do you think people should be able to own guns to protect themselves?” before asking a question about gun control will decrease the response in favor of gun control.

All political issues are affected by how the question is asked. Many states, including California, Texas and New Jersey, have procedures for putting issues on the ballots for voters. Every time there is a struggle over how the question is worded. A slight change in style can mean the difference between its passage or rejection, even though the substance does not change.

Recently supporters of gay marriage published a poll claiming that most people favored it. This was in spite of the fact that every time the issue has been put to voters, gay marriage has been defeated. The trick to the poll? It used a word that has an ambiguous meaning: “sanction”. “Sanction” can mean either to endorse something, or to penalize it. The poll asked whether gay marriage should be “sanctioned”. The confusion created was probably intentional. Always check the wording of polls to see if they are clear and unambiguous.

Another trick in manipulating polls is to ask a general question rather than a specific one. “Do you support equal rights for women?” is likely to generate a favorable response. However, voters repeatedly defeated referenda for the “Equal Rights Amendment,” which would impose burdens like the military draft on women. An indication of whether people support the Equal Rights Amendment would be to ask the question: “Do you support drafting women into combat?” That is not likely to generate a favorable response.

Before interpreting a poll, always look at the details of the questions that are asked.

Unavoidable Flaws in Polls

While some of the flaws in polls are due to manipulation, other flaws are unavoidable but are often ignored. What are some examples?

Polls only capture public opinion at a given time. They are typically useless in predicting opinion in the future.

At one time the people wanted to make Jesus Christ their king, and then a few weeks later the people wanted to crucify him. And Jesus knew that would happen.

President Reagan never had very high approval ratings as president. Yet his approval rating when he left office after his second term was higher than any other president on his last day of office. Sometimes what is unpopular at one given time becomes more popular later. An unpopular teacher may be one who assigns a lot of homework. However, he may become very popular later when students do well based on what they learned!

The first President Bush broke records for high approval ratings during the Gulf War in 1991. Yet by Election Day in 1992, his approval rating was very low and he was defeated by a wide margin by President Clinton.

Increasing government spending can be popular at first, because voters receive additional monies from the government. But as the effects of a larger deficit are felt, unpopularity can set in.

In the eyes of the public, a person can be a hero today but a goat tomorrow. Reading public opinion polls requires anticipating how the public will feel in the future. That is very difficult, often impossible, to do.

Polls also fail to capture the strength of people’s support. A “yes” or “no” answer to an opinion question does not reveal how strongly one feels about their answer. A “yes, but I don’t really care much” is very different from a “yes, and I’ll volunteer my time to make sure this happens.” Many politicians have been fooled by public opinion polls about abortion. Many against abortion are willing to volunteer their time and effort to stop it, and often vote on that issue alone. But many of those who accept abortion do not vote based on that issue.

It is not only the opinion of the public that matters in political races. The opinion of the media is also very influential. The media is pro-abortion, and will give less favorable treatment to a candidate who is pro-life.

Often polls will be used to portray a candidate as a “winner” to the public. Some voters like to back someone they think will win, simply to be in the winning side. This is a bigger problem in foreign countries like England, where “underdogs” are unpopular. It is a factor in the United States with respect to raising money, because popularity in polls makes it easier to raise campaign money. So candidates will employ their own pollsters to generate as positive data as possible about their candidate.

The Political Spectrum

What do political polls measure? The many different political views of Americans, which span every possible part of the political spectrum.

Conservative, liberal, libertarian, green (environmental), prohibition, socialist, communist – you name it, and there are thousands or millions of Americans who support a given political view.

It is not easy to create an accurate model for all the political views. In an oversimplification, the media will pretend that politics is on a straight line from left (“liberal”) to right (“conservative”). Under this model, everyone can be placed at some point on the line. In the middle are the “moderates” or “centrists”, which is how many in the media view themselves. Under this model, the two ends are “extremes”, and people there are pejoratively called “extremists”.

This linear model of politics affects the thinking of many. It favors candidates who lack clear positions or principles and who are willing to “go with the flow.” The flow, in turn, is dictated by the media. So this is a model that favors media influence. Also, the term “moderate” has positive connotations, so people who view themselves as moderates prefer this model.

It also makes it easier to report on politicians. Those who do not take strong positions are simply called “moderate”, while others are painted as “extremist”depending on how clear their positions are. Moderates tend to favor compromise. Presidential moderates, for example, like to sign treaties to resolve disputes.

Is the model accurate? Often it is disproven in elections. Jimmy Carter and Bill Clinton were both liberal presidents, but many conservatives voted for them. If politics is liberal-moderate-conservative, then a liberal candidate should never receive votes by conservatives. So how did that happen?

Perhaps politics is not correctly modeled by a straight line. Another way to view it would be as a circle. When you go very far to one side, you come out on the other side. The very liberal and the very conservative agree with each other on some issues more than with the middle.

“Free trade” is an example. Liberals and conservatives mostly oppose free trade with other countries, while moderates support it. War is another example. Liberals and conservatives often oppose foreign wars, while moderates almost always support it. The two political ends agree with each other more than with the middle.

Can you think of more examples? How about the power of the federal government in criminal cases? Liberals and many conservatives tend to oppose increased federal prosecutorial power, while moderates support it.

In an election, it depends on which issues are most important on Election Day to see whether moderates or liberals will vote more like conservatives. If the big issues are free trade and foreign alliances, then the election may be liberals and conservatives against the moderates.

On the Supreme Court, cases will sometimes wind up with the liberals and conservatives voting for one side, and the moderates voting for the other.

Religious and Regional Influence on Votes

What is the best single predictor of someone’s vote? How often they attend church or house of worship. In general, the more a group prays, the more conservative the group votes. (Of course, there are many individual exceptions.)

This may be one reason why prayer has become such a political issue. Public prayer has an influence on which side wins elections. The United States is the most conservative large country in the world because its people are the most religious.

Every candidate is touting their religious beliefs near Election Day. President Clinton liked to be photographed leaving church hand-in-hand with Hillary Clinton. Their actions after the election would be a different matter, of course.

Another big influence on people’s vote is regional politics. The United States is geographically large. When the North goes one way, the South usually goes the other. Right now, the North is trending Democratic, which makes the South trend Republican. 150 years ago, the North went Republican and the South went Democratic.

This regional split makes it very difficult for either party to grab control of the federal government. It ensures divided government for many years to come.


1. Should there be limits on individual campaign contributions?

2. Who do you think will win the presidency in 2004?

Lecture 6 Quiz