Dick Morris
The perils of polling
Polling is in trouble. Big trouble. The data is no longer reliable and won't be for several more years. Why were all the national polls (except Zogby) wrong in predicting that George Bush would get more votes than Al Gore in 2000? Why were all the polls wrong again in predicting a close Senate race between Rep. Rick Lazio (R) and Hillary Clinton in New York that same year? The telephone poll is no longer a credible method of measuring public opinion. In 28 states, the state legislatures have passed laws giving telephone users the right to opt out of receiving telemarketing phone calls, including public opinion surveys. More and more voters are availing themselves of this right and the pickings for telephone polling firms are getting more and more scarce.
In Connecticut, for example, 29 percent of the state's households have chosen to use the opt-out and these 500,000 people cannot be contacted by America's polling organizations. Five percent of Connecticut households join the ranks of those refusing to take telemarketing calls each year.
Even beyond the formal opt-out which makes it illegal to call certain voters when taking public opinion polls, the "hang up" factor is looming larger and larger in telephone polling. The anger which leads almost one in three of Connecticut's voters to refuse to take marketing or polling calls exists throughout the land and further reduces the ability of phone surveys to amass a statistically valid sample.
So telephone polling is back where it was in the days of the Literary Digest poll of 1936, which famously predicted an Alf Landon victory over Franklin Roosevelt because of the skewed demographics of a telephone poll.
Especially in off-year elections, the voters who refuse to answer telephone polls are just the ones who will most likely come out and vote on Election Day. Upscale, aware of their rights, and determined to act to protect their privacy, these are the same people who vote when the percentage of voting age population that casts its ballots drops to 35 percent during off-year contests.
The problem is that there is no substitute that is any better. In-person polling has not been feasible for many years. Whatever tales the polling firms tell, their interviewers are not the sort who will willingly risk their lives by canvassing ghetto apartment buildings in the evening. Besides, the cost and time required for personal interviews makes it prohibitive for a political contest.
Internet polling is growing more reliable every day, except for its blind spot - the 40 percent of Americans who do not go online.
According to the Census Bureau at the start of 2002, 60 percent of whites used the Internet, 40 percent of blacks and 34 percent of Hispanics. Interestingly, 25 percent of the poorest fifth of the nation used the Internet.
As the opt outs from telephone polling increase and Internet use continues to grow by about 8 percent each year, Internet surveys will become more accurate than telephone interviewing. But, now, the number of opt outs and the proportion of the voters who do not go online are about the same in many states. Of course, the voters who refuse phone polls are more likely to vote than those who won't use the Internet, but each medium has its own blind spot at the moment.
The other problem with Internet polling is that it is difficult to get a statewide list of e-names. An essential premise of polling is that each voter must have an equal opportunity to participate. But with opt-outs on the one hand and the limited availability of e-names on the other, this requirement is a practical impossibility in current American life.
Many will find the difficulty in polling something about which they do not get too upset. But we need to take all polls conducted in this atmosphere with a healthy dose of skepticism. The pinpoint accuracy that was possible in earlier days is no longer achievable. Politicians and pollsters must admit the new shortcomings of their once highly reliable polls.