WASHINGTON - Call waiting. Caller ID. Answering machines. Computer modems. Cell phones. Home fax machines. Voice mail. Pushy telemarketers.
For most of us, they are conveniences or annoyances of modern life. For opinion pollsters, they can add up to big trouble. As the numbers and varieties of polls spiral, it's growing harder to get people to answer their questions, and harder to get accurate results.
''Response rates are not what they used to be,'' said pollster John Zogby, referring to the willingness of those called to participate in polls. ''When I started in this business in 1984, they were averaging around 65 percent. Now we are down around 33 or 35 percent.''
Indeed, changes in technology and the attitude of Americans have lent an unexpected wrinkle to Campaign 2000: The science of measuring public opinion has come in for increased scrutiny and debate.
In recent weeks, several well-known polls have reported questionable shifts in public opinion, leaving pollsters self-critical and defensive and prompting what GOP pollster Ed Goeas called ''the debate over polling.'' At least two major polling operations have had to adjust their techniques.
The controversy started in mid-September, when three respected polling operations announced dramatically different views of the presidential race.
A Sept. 15 Newsweek poll gave Vice President Al Gore a 14-point lead, while the bipartisan Voter.com-Battleground poll had Governor George W. Bush ahead of his rival by between 2 and 5 points. The CNN/USA Today/Gallup nightly tracking poll, meanwhile, was somewhere in the middle.
Newsweek subsequently adjusted its methodology by adding to the size of the polling sample. And Goeas, who conducts the Voter.com-Battleground poll with Democratic pollster Celinda Lake, tinkered with the order of its questions, and acknowledged that they may have been using a sampling method more suited for the final days of the campaign, when opinions are firm, than for mid-September, when the situation is more fluid.
The flap over the campaign polls is more than just inside baseball. It is symptomatic of a larger phenomenon that could affect the American political, media, and marketing worlds in the years ahead.
At stake is the validity of polling as we know it, in which polling firms contact 250 to 500 respondents by telephone on a given night and ask them about their voting, shopping, or entertainment habits. Candidates, corporations, and advertising agencies all rely on such random sampling to tell them what the public thinks.
Part of the challenge is technical: The increased number of voters who have caller ID, computer modems, and similar devices in their homes has made the polling business more complex, costly, and difficult than ever. It used to take Republican pollster Frank Luntz three phone calls, on average, to reach a respondent. Now it takes six.
But there have been attitudinal changes as well. At the same time that they are tying up their phone lines with new equipment, Americans are showing growing impatience with intrusive telemarketers who want to sell them on long distance service, a hot stock, or a local charity. A call from a polling organization is not much more welcome.
Polls are no longer a novelty. People who might once have been flattered to be interviewed by a big news organization now see it as an imposition, especially at the end of a long, busy workday.
And so pollsters are getting more busy signals, recorded messages, and abrupt (and sometimes rude) goodbyes.
Regional differences further complicate the problem. A Midwestern state like Missouri or Iowa, for example, is a pollster's dream: Folks get in from the farm or factory at suppertime, join their families for dinner, and show a native politeness to callers.
Texas is more difficult, notorious for its huge number of telephone answering machines. And New York is a nightmare. Known for being blunt if not downright surly, many New Yorkers work past dark, have a long commute, and arrive home later each night, giving pollsters a narrow window to catch them before bedtime. On the three coasts, the number of non-English-speaking people only increases the difficulty of building a good sample.
As they grapple with such changes, pollsters are arguing about methodology. Democratic pollster Peter Hart, for instance, surveys registered voters until the final weeks of a campaign, when interest in the race peaks and he can better determine who are the highly-prized ''likely voters.''
Goeas, on the other hand, thinks it is better to screen for the ''likely voters'' earlier in the race, particularly by discounting younger voters, who tend not to follow through on their intent to vote, and people who register in ''motor-voter'' drives, then don't show up on Election Day.
Though 60 percent of young voters tell pollsters that they intend to vote, only 33 percent cast a ballot, said Goeas. ''By allowing more of those young voters into your sample you are going to see more volatility.''
Goeas prefers to poll only on weeknights, because busy suburban families are more likely to be out of the house on weekends, tilting weekend results toward the Democrats.
''On weekends you get a more Democratic, blue-collar sample. You are under-reporting Republicans, especially married voters,'' said Goeas. ''Quite frankly, a mother who will take 20 minutes on a weekend to talk to a pollster is not a normal person.''
When Zogby took the temperature of the New York Senate race last weekend, he found Hillary Clinton leading Representative Rick Lazio 48 to 44 percent with a whopping 71 to 23 percent advantage among an important minority that always turns out at the polls: New York's Jewish voters.
But Goeas challenged the accuracy of the poll because Zogby did his polling over the Rosh Hashana weekend, the Jewish New Year. The Republican pollster claimed that the most Orthodox Jewish voters - cultural conservatives who were more likely to vote Republican - were observing the High Holy Days and unlikely to respond to a pollster's phone calls.
Because Orthodox Jews were underrepresented, Goeas asserted, the Zogby results must be skewed, and Lazio is in better shape than he seems.
Not so, said Zogby: His firm had made adjustments in its methodology to correct for Rosh Hashana. It is Goeas whose polls are out of kilter for refusing to poll on weekends, when Democrats are more likely to be home, Zogby said.
''I've also heard you don't poll in East Texas on Wednesdays because everyone is at the Assembly of God Church. And that you don't call Manhattan in July because everyone is in the Hamptons,'' Zogby said, dismissing Goeas's theory with a chuckle. ''I put not polling on Rosh Hashana in that category.''
Given time and money, said Zogby, pollsters have shown great skill at compensating for regional, technical, and attitudinal changes.
''Has it made it impossible to do our job? No,'' he said. ''You can overcome these difficulties. We are not in a crisis situation.''
Still, not every client has time and money to spend on a thorough survey. To feed the voracious demand of the media, some polling firms are rushing into the field on a tight budget, relying on smaller or poorly screened samples and sacrificing accuracy, said Luntz, the GOP pollster.
The media, always looking for the most dramatic results, then compound the problem by hyping the polls that show the widest margins or shifts of opinion, instead of averaging the various polls to find a norm.
''The use of polls in this election cycle is completely and totally out of control,'' Luntz said.
The closeness of this year's presidential race, says Hart, has served to magnify the differences among the polls.
Even the best poll has a 3- to 5-point margin of error, Hart noted. So a poll giving Bush a 2- or 3-point lead can be just as accurate as another poll giving Gore a 2- or 3-point lead the very same night. Declaring that either candidate has seized momentum on the basis of one night's tracking is foolhardy, said Hart, who warned the public: ''Don't fall into that trap.''
Goeas noted there were also wide disparities among the polls during the 1996 campaign, but they weren't as noticeable because they all agreed on the overall trend: Bill Clinton was running well ahead of Republican Bob Dole.
In the 2000 campaign, the trend is that Gore and Bush are running neck and neck. A swing of a few points can put one or another in the lead - changing the face of the race - and so differences are magnified.
The good news is that, as voters reach decisions and pollsters identify and question ''likely'' voters, ''it all becomes more reliable as you move toward the election,'' Goeas said.
