For most of American history, when a politician wanted to know what voters were thinking, he would call or write to local political leaders to find out what they were hearing at the grass roots. Journalists would go out into the streets, to community events, or perhaps just to a local watering hole and actually talk to average people about what they were thinking.

Oftentimes, it was impossible to figure out what people thought about specific issues. They might be highly technical, complicated or just too boring for anyone outside the Capitol to care about. In these cases, members of Congress would have to use their judgment, often based on a political philosophy, in order to decide what to do.

Today, this form of politics has pretty much gone out the window. Politicians are too busy raising money and journalists are too busy faking documents to bother with talking to real people. Instead, they rely on public opinion polls and focus groups arranged by professional pollsters to tell them what people are thinking. Gone are the days when a politician would figure out for himself how to vote. Today, all he wants to know is which way is the wind blowing.

Senator Kerry is a perfect example of the danger of this approach. On the day he had to cast a vote for the Iraq invasion, the wind was blowing in a pro-war direction, so he voted “yea.” Then later, when the wind had reversed course, he voted to deny funding for the troops in Iraq. Thus he was simultaneously pro-war and anti-war.

Mr. Kerry would have been far better off sticking with one side or the other, instead of trying to split the difference. In the pre-poll age, perhaps he would have. He would have been forced to make up his mind and take a stand. But with the easy availability of polls telling him what voters wanted, the pressure to be on the popular side of the war was irresistible, leading to his Iraq flip-flop.

There’s nothing wrong with a politician trying to figure out the way his constituents want him to vote. He is, after all, their representative, casting a proxy for how they presumably would vote themselves on a given issue if we had a pure democracy. The problem is that the public opinion is not fixed. It changes, sometimes radically in a short period of time. Hence, the politician who devotes himself to following public opinion is necessarily going to be extremely inconsistent – something voters don’t usually appreciate.

Unfortunately, many politicians spend their careers chasing the will-o-the-wisp of public opinion in order to determine their positions, at the expense of analysis, thought, and consistency. And they frequently use polls to figure out which way the political winds are blowing.

Consequently, advocates now use polls to advance their agendas. If polls were truly scientific, if the public were well informed, and if public opinion was stable, this might help advance political debate. However, none of those things are true. Moreover, it is too easy to load questions so as to get pretty much whatever answer is wanted by whoever is paying for the poll.

Until recently, this wasn’t that much of a problem. National polls were very expensive and only conducted by reputable organizations. But in recent years, computers and the sharp decline in telephone charges have greatly lowered the cost of polling and increased the number of polling companies. Now, just about any special interest group can afford to do a poll showing overwhelming support for its position. And presidential campaigns can afford to poll almost continuously.

Thus much of the volatility we have seen in the presidential race this year simply stems from a vast proliferation of polls. There is a poll from some major news organization almost daily.

But bias also plays a role in volatility. For example, it is well known that Republicans tend to vote in higher percentages than Democrats. Thus any poll based on the general population is going to tilt toward the Democrats. Narrowing the population down to registered voters will improve Republicans prospects, and polling only likely voters will improve them still more.

News organizations also use various tricks – I mean “adjustments” – that have the effect of tilting their polls in a Democratic direction. They do this by treating political affiliation as if it is something fixed and immutable, like sex or age. So if there is movement toward one party, the poll is weighted so as to eliminate this factor from the results. With the Democratic Party having steadily lost its dominant position over the last 20 years, the effect of this weighting procedure is to give Democrats more influence on polls.

Fortunately, the only poll that matters is the one on Election Day.

Mr. Bartlett is a senior fellow at the National Center for Policy Analysis.