With only 48 hours left to go in a riveting presidential race, anyone with an interest in politics is desperate for the latest trends from battleground states. No need to prolong the suspense: As of last night, President Bush was up by one point over Democrat John F. Kerry in Florida and Ohio, down by two points in Pennsylvania.

Who says? That would be Scott Rasmussen, who is following this anxious election with contentment from his home town of Ocean Grove, N.J., confident that he is divining the mysteries of democracy with the help of a computerized phone bank in Texas and several pleasant-voiced women in the Midwest.

Rasmussen is a model of a proliferating -- and, in some quarters, controversial -- new breed in politics: the entrepreneurial pollster.

Traditionally, most political pollsters had party affiliations, generally working for one candidate or the other, or else polled on behalf of news media organizations or think tanks. Rasmussen -- who as a young man growing up in Connecticut joined with his father as a founder of ESPN -- a few years back saw a new polling niche to be filled by feeding an insatiable appetite among political junkies in the general public. His political polls are not conducted for clients; they are conducted for himself, with publicity and profits as the motive.

"We serve a different audience," said Rasmussen, 48, in a telephone interview. "We are here to make some money."

His business model is organized around a paradox of American politics: Although voters often bemoan polls and the politicians who rely excessively on them, lots of ordinary Americans delight in the data they produce. Rasmussen's Web site, www.rasmussenreports.com, was getting 1.3 million hits a day last week from people who wanted to know the latest about his daily national tracking poll, and the number is rising each day. By Election Day, Rasmussen said, he's is hoping for 2 million hits a day.

That's on the home page. For full access to the site, one must be a "premium member," which costs $95 annually and offers more detail on state-by-state tracking polls as well as Rasmussen's analysis. Five thousand people, he said, have signed up. As expensive as this is, it's a bargain compared with full access to pollster John Zogby's site, which costs $140 for the highest level of membership. That gives members access to the specifics of a poll's questions and responses broken down by categories, known by political insiders as the "cross tabs."

Rasmussen in particular stirs controversy because of his polling methods. Unlike traditional pollsters, he does not have a room full of questioners phoning randomly around the country. Instead a computerized voice, using an automated calling system operated by a Texas firm, conducts the interviews. People on the other end of the line who are willing to cooperate give answers by punching buttons on their telephone keypads.

Another firm, New York-based SurveyUSA, also uses the disputed method in the dozens of polls it conducts for local television stations and makes available on its Web site.

Some academic experts on public opinion research believe such "robo calls" are methodologically unsound. By this reckoning, people are less likely to give thoughtful answers to a computer. Some mischief-makers, critics believe, might even be more prone to give deliberately false answers.

"I would not rely at all on the Rasmussen data," said Michael W. Traugott, a political scientist at the University of Michigan who has written academic articles criticizing robo-calling. "The computer does not know who's really at the other end of the phone."

Rasmussen counters that robo-calling has some advantages. Each interview is the same, so no one can be affected by minor differences in wording or voice inflection. He uses voices of women in their thirties from midwestern states -- because his research has found they have the most pleasant, neutral-sounding voices -- to record questions. The voices sound so open and inviting, he said, that many people do not realize right away that they are not talking to a real person.

As far as reliability goes, Rasmussen said his results are sound. He acknowledges that, like most pollsters, he's had hits and misses. In 2001, his data proved generally accurate for governors races in New Jersey and Virginia. In 2000, by contrast, he had what he called his "Dewey Defeats Truman" moment when Rasmussen tracking polls showed Republican George W. Bush widening his lead over Democrat Al Gore in the closing days, when as Election Day showed, the opposite was occurring.

Rasmussen knows he has only hours left in which many eyes will be on him -- at least for this election cycle -- but is enjoying the last moments of drama. "This election has been so close for so long, it's almost reached a frenzy," he said. "My only interest in this election is that if our final tracking poll says Bush is up by two points, he wins by two points, and if it says Kerry is ahead by two points, then Kerry wins by two points."