According to the latest poll ... These days, it's hard to turn on a TV or open a newspaper without seeing a poll. But not all polls are created equal, and it can be tricky figuring out what they really mean--if they mean anything at all.

A hypothetical public-opinion poll shows President George W. Bush Senator John Kerry by a margin of 48 to 43 percent. A poll the next month shows Kerry leading Bush by 47 to 44. Has there been a big shift in how people plan to vote?

Maybe. And maybe there's been no shift at all.

With the November elections fast approaching, it's hard to read a newspaper or watch the news without seeing polling figures that try to handicap the presidential race. Not only are people fascinated by the opinions of their fellow Americans, but politicians use polls to help tailor their messages to sway voters, and journalists use them to find out what people care about and why.

But it's important to know the limits of what a public-opinion survey can reveal.

A SNAPSHOT, NOT A CRYSTAL BALL

A poll tells us about the present, not the future. It's not a crystal ball; it's a snapshot of how the public is thinking at a particular moment, and not a perfectly sharp snapshot at that. Although a large portion of the American public is committed to one political party or the other and holds firm positions on a wide range of issues, there is always a segment that wavers, or stays unconvinced (or uninterested) until late in a political campaign. Those "swing voters" ultimately push elections back and forth, generally between the Republicans and the Democrats, and their views can keep changing up to the very last minute.

One of the most embarrassing errors in polling history came in 1948, when the major polling organizations all declared that Thomas Dewey, the Republican Governor of New York, would beat Democrat Harry Truman in the presidential election. All of them had stopped interviewing the public several weeks before the election and missed a late swing toward Truman. In addition, a study after the election found their samples had too many middle-income and wealthy people, who were more likely to vote Republican.

As late as election night, many political experts were convinced Dewey was going to win and the photograph of a smiling Truman holding up a copy of the Chicago Daily Tribune that wrongly declares "Dewey Defeats Truman" has become a political and polling icon.

POLLING PITFALLS

Today, surveys are conducted right up to Election Day and to the polling booth itself, using a method called "exit polling" in which voters are interviewed at polling places just after they have cast their ballots.

But even exit polls have pitfalls. In Florida, in the 2000 presidential race, a combination of bad polling data in a handful of precincts, compounded by mistakes in the actual vote counts, led most of the television networks to call a race first for Democrat Al Gore, and then for Republican George W. Bush--that was in fact too close to call. Florida ended up determining the election's outcome, which was ultimately decided by the Supreme Court.

HOW POLLS WORK

No poll is perfect--there are errors built into the process itself. The basic premise behind public-opinion polling is that by asking the opinions of a surprisingly small number of people, you can get a good sense of what an entire city or state or country is thinking. It's similar to the idea that you can tell if a pot of soup is too salty by trying a few spoonfuls, without having to eat the whole pot.

Most reputable political polling today relies on a method called "probability sampling." The idea is that if you select people at random from a whole population, no matter how large or small, you have a good chance of reflecting, within a few percentage points, the opinions of everyone.

But those few percentage points can make a difference. A survey of 1,000 people--which is a typical number for a news-media survey--will have what is known as a "margin of sampling error" of plus or minus 3 percentage points, 95 percent of the time.

The margin of sampling error is one of the most important things to know when trying to understand the significance of a poll: A 3-point margin means that if 48 percent of the people interviewed said they would vote for Bush, the real number--the one you would get if you asked everyone in America--could be as high as 51 percent or as low as 45 percent, 19 times out of 20. (And one time out of 20, the error could be even larger.)

So with a 3-point margin of error, Bush's 48 percent in the example at the beginning of this article could also be 45 percent, and Kerry's 43 percent could he 46 percent, putting Kerry ahead. This example illustrates one of the most common mistakes people make in reading polls: thinking of the percentages as precise points, when they're actually more like fuzzy blobs. That's why polls are least helpful in figuring out who is ahead when an election is close, as the presidential election has seemed to be this year.

There are other possible sources of distortion as well. Questions must be worded carefully so that they are understood the same way by different types of people in different parts of the country and so that the questions themselves don't sway the results. Even the order in which questions are asked can unintentionally affect the way people answer.

Opinions on a particular issue can also shift rapidly as a result of recent events: Feelings about whether the U.S. was right to have invaded Iraq, for example, might be influenced by a run of particularly grim or hopeful news from Baghdad.

Studies have found that people largely tell the truth to pollsters, although the truth they tell is based on their understanding of events, which is not always perfect. Sometimes they'll give an opinion about something they didn't even have an opinion about until the pollster called. And sometimes social pressures or other factors will lead people to fib. For example, more people report that they are registered to vote than actually are, probably because voting is considered a good thing to do. And after a presidential election, a larger percentage of the public almost always reports having voted for the winner than actually did.

INSTANT EQUALS NONSENSE

Beware of poll pretenders, such as "instant" TV polls and online polls in which people are asked to give opinions about a topic of the day. Without organized sampling, such polls don't represent the opinions of anyone but those who happened by and answered the question--and maybe the friends they told about it.

"If the poll just interviews people who volunteer for the poll, like most Web polls, ignore it," says Warren Mitofsky, a polling expert who has influenced the field for decades. "Results cannot he generalized to some larger group."

As Internet use has become more widespread, some reputable polling firms have begun working on ways to conduct legitimate surveys online, though most surveys are still conducted by telephone. Pollsters have long had to adapt to changes in technology. When scientific surveys began in the mid-1930s, most polling was done face-to-face.

CALLING COMPLICATIONS

One of the new challenges facing pollsters is the growing number of people who have only cellphones: Census estimates from this year show that 5 to 6 percent of the U.S. population have given up their land lines, and pollsters generally do not call people on cellphones because of legal and other issues. Do people who have land lines have identifiably different opinions from those who don't? Researchers are studying that now. Another continuing issue of concern for pollsters is the number of people who refuse to respond to surveys or can't be reached. How do they differ from the people who will answer the phone and take the 15 or 20 minutes that most surveys require?

Despite all these cautions, however, properly conducted polls are still the best way for us to find out what people are thinking at a particular moment in time, and in a democracy, that's important to know. Be skeptical when candidates say they don't pay attention to polls, or that "the only poll that matters is the one taken in the voting booth." Very often they're the ones who think they're behind.

Looks Can Be Deceiving

          June 23-27     July 11-14

KERRY         42             45
BUSH          43             42
NADER          5              5

Margin of sampling error +/- 3%

In these recent polls, Senator Kerry appears
to gain over President Bush (with Ralph Nader
trailing). But since the polls both have a
3 percent margin of error, neither candidate
could rightly claim to be ahead.

SOURCE: New York Times/CBS News polls

Note: Table made from bar graph.

TEACHING OBJECTIVES

To help students understand how polls work, why they seem to be everywhere, especially in an election season, and how to figure out what they mean and don't mean.

CRITICAL THINKING/MORAL QUANDARY:

Ask students to pretend they are running for the city council or some other public office in their community. They are against a proposed highway extension because environmentalists say it will harm wetlands. But opinion polls show growing public support for the extension. Which course do your "politicians" take? Do they shift their stand on the extension to accommodate public opinion? Or do they hold firm and try to shift opinion against the extension?

Students should present arguments supporting both strategies. (Possible "Pro" extension: The people's opinion should be the foundation of public policy. Possible "Con" extension: Part of a political leader's job is to help educate citizens on important public-policy issues.)

POLL WATCH: Between now and Election Day, have students monitor the poll ratings of President Bush and Senator Kerry. Keep a log, noting general poll ratings and ratings on specific issues. After the election, compare the percentage of the vote won by Bush and by Kerry with earlier poll tallies. How close did the pollsters' earlier findings come to the final vote?

DISCUSSION QUESTIONS

* Do you agree that people are fascinated by the opinions of others? If so, why do you think this is so?

* Why do you think some people refuse to respond to pollsters?

FAST FACTS: One of the first presidential polls was conducted in 1824. Two newspaper polls picked Andrew Jackson, who won a plurality of popular and electoral votes. But no candidate won a majority, so the election was decided by the House of Representatives, as provided for in the Constitution. John Quincy Adams won.

In the mid-1930s, George Gallup, Elmo Roper, and Archibald Crossley developed modern polling techniques.

WEB WATCH: www.pollingreport.com is an independent, nonpartisan collection of polls on the presidential race and other public issues.

Polling Americans About Their Presidents

Public-opinion polls will be in the news almost every day as Election Day approaches. Politicians want to know who's leading in the race for the White House because the results will affect how and where they target their campaigns.

But there is one kind of presidential opinion poll that has little to do with the campaign and more to do with Americans' sense of history. These are the polls in which people are asked to look back and compare past Presidents with the current Chief Executive. The data in this graph, from a June Gallup poll, show how Americans rated six Presidents, including George W. Bush.

Remember, a poll is a snapshot in time. If the same poll were taken today, the results might be different because people's opinions can change as they gain access to new information, or because they have rethought their earlier views of Presidents. Use the data in the graph to answer the questions to the right.

Note: Totals may not add up to 100 percent because "no opinion" responses are not included.

1. Which President had an Outstanding rating that was almost exactly half as high as Bill Clinton's Average rating?

--

2. Two Presidents have identical. Outstanding ratings. Who are they?

--

and

--

3. Two Presidents have identical Average ratings. Name them and their scores. Then name the President whose score is one point lower--and his score.

President -- score -- President -- score -- President -- score --

4. President George H.W. Bush (1989-1993), father of the current President, is not shown on the graph. But he received an Average rating equal to the highest Above Average rating listed on the graph. What was George H. W. Bush's Average rating?

--

5. How many percentage points separate the President with the highest Below Average rating and the one with the lowest Below Average rating?

(a) 30 (c) 2 (b) 23 (d) 6

6. CRITICAL THINKING/RESEARCH ASSIGNMENT: Research any of the Presidents listed on the graph and write a brief note in which you grade the President as good, or not-so-good.

--

QUIZ 2

Polling America

1. Some critics want a ban on news of political-poll results within a certain number of days before an election. What do you think their argument is?

--

2. One reason polls can miss shifts in support for political candidates is the changing views of

a uncertain Democrats.

b uncertain Republicans.

c centrists.

d swing voters.

3. "Exit polling" refers to opinion surveys conducted

a after a political leader leaves office.

b after voters leave their polling place.

c as voters are on their way to the polling place.

d before the polling places open.

4. "Probability sampling" refers to

a the process by which poll questions are written.

b the likelihood that any particular person will be surveyed.

c how a randomly selected number of people is surveyed to measure the opinion of a larger group.

d ways to measure the accuracy of responses.

5. Why are Internet polls notoriously unreliable?

--

6. A relatively new product is adversely affecting the ability of pollsters to contact potential interviewees. This product is the

a cellphone.

b home computer.

c push-button phone.

d pager.

Answer Key

Graph Exercise * page TE 4

1. Ronald Reagan

2. John F. Kennedy; Franklin D. Roosevelt.

3. George W. Bush, 30; Bill Clinton, 31 ; Ronald Reagan, 31.

4. 55 percent.

5. (a) 30.

6. Answers will vary.

Richard J. Meislin is the editor of news surveys and election analysis for The New York Times.