||One of the nation's top political analysts, Stuart Rothenberg, dissects politics at the congressional and statewide levels.|
Stuart Rothenberg: Reading between the polls' lines
WASHINGTON (CNN) -- A group of at least nine national polls of the 2000 presidential race have come out so far in June, and the differences are so significant that they have to make any sane person wonder about the reliability of public opinion surveys.
My advice about this is pretty simple: Don't place your faith in any single poll number. Remember, every poll has a margin of error. Instead, look at all of the polls, keep in mind who and what they are measuring, and try to get a general sense about what the surveys are finding.
The current presidential polls range from a tight 42 percent-40 percent George W. Bush advantage over Al Gore (with Ralph Nader at 3 percent and Pat Buchanan at 2 percent) in a June 22-23 Princeton Survey Research Associates poll of 750 adults for Newsweek to a 13-point Bush lead in the most recent CNN/USA Today/Gallup survey. The CNN poll of 560 likely voters was conducted June 23-25.
Here's a look at the other polls, which include some ballot tests of just the two majority party candidates, and some tests of Gore, Bush, Nader and Buchanan:
--Research 2000, June 5-8, 821 likely voters: Bush over Gore 45 percent-43 percent.
--Fox News/Opinion Dynamics, June 7-8, 900 registered voters: Bush over Gore 46 percent-38 percent. Bush 41 percent, Gore 38 percent, Nader 4 percent, Buchanan 2 percent.
--Princeton Survey Research Associates for Bloomberg News, 1,208 adults: Bush 41 percent, Gore 38 percent, Nader 4 percent, Buchanan 2 percent. 924 registered voters: Bush 44 percent, Gore 40 percent, Nader 4 percent, Buchanan 3 percent.
--Los Angeles Times, June 8-11, 1,686 registered voters: Bush over Gore 50 percent-40 percent.
--ABC News/Washington Post, June 8-11, 1,204 adults: Bush over Gore 49 percent-44 percent. Bush 44 percent, Gore 42 percent, Buchanan 4 percent, Nader 3 percent. 966 registered voters: Bush over Gore 49 percent-45 percent. Bush 47 percent, Gore 41 percent, Buchanan 4 percent, Nader 4 percent.
--Voter.Com Battleground Survey, June 11-13, 1,000 likely voters: Bush over Gore 52 percent-40 percent.
--NBC News/Wall Street Journal, June 14-18, 2,010 adults: Bush over Gore 49 percent-41 percent. Bush 43 percent, Gore 38 percent, Nader 7 percent, Buchanan 4 percent
--Zogby, June 16-20, 1,011 likely voters: Bush over Gore 46 percent-39 percent. Bush 42 percent-Gore 36 percent, Nader 4 percent, Buchanan 3 percent.
--Princeton Survey Research Associates for Newsweek, June 22-23, 620 registered voters: Bush and Gore tied at 45 percent. Bush 42 percent, Gore 40 percent, Nader 3 percent, Buchanan 2 percent.
--CNN/USA Today/Gallup, June 23-25, 560 likely voters: Bush over Gore 52 percent-39 percent. Bush 50 percent, Gore 38 percent, Nader 6 percent, Buchanan 2 percent.
In two-way ballots, the results show anything from Bush and Gore in a statistical dead heat to the Texas governor holding a solid double-digit margin over Al Gore. In three of them, Bush is at or above the critical 50 percent mark. In a couple of others, he's at 49 percent, and in two, he's in the mid-40s.
In other words, all of these surveys can't be right. At this point, I look for trends among the differing polls to see what I can learn.
Six of the surveys did both two-way Bush versus Gore tests, as well as four-way ballots with Green Party nominee Ralph Nader and likely Reform Party nominee Pat Buchanan included. In five of those six, Bush performed better in the two-way ballot tests, while in one -- the Newsweek survey -- Bush did better in the four-way.
The two polls conducted by Princeton Survey Research Associates (for Bloomberg and Newsweek) tended to show a closer race than other surveys taken during the same general time period. That's worth keeping in mind when Princeton reports other survey results in the future.
Anyone who looks for "movement" in the polls by comparing a Princeton Survey Research poll with a CNN or Wall Street Journal survey, for example, is looking for trouble.
Never -- and I mean NEVER -- compare two polls by different media outlets to find a "trend" in public opinion. It's okay to compare the Princeton surveys with each other to try to identify movement over time, but it is definitely NOT okay to compare polls that have different samples (registered voters versus adults, for example) or those conducted by different media outlets.
Some talking head experts violate those rules of polling analysis, of course, but that doesn't mean you should.
Virtually all of the polls show third party hopefuls Ralph Nader (Green) and Pat Buchanan (Reform) in the low to mid single digits. That leads me to believe that that's exactly where the two candidates are at this point. Of course, those results don't tell us where those hopefuls will end up in November. But they point out that both men have a long, long way to go in order to get into the presidential debates.
After looking at all of the results, I think it's pretty safe to say that George W. Bush is ahead of the vice president. My own guess is that he has a lead of about 6-9 points -- somewhere between the extremes of all of the surveys I have mentioned. That's a good position for him to be in at this point -- a very good position, considering the state of the economy and the fact that Gore is so well-known.
Luckily, I don't have to be too precise at this point in the election cycle. But for the Bush and Gore campaigns, the numbers matter. They are likely to employ different strategies if the race is really a three-point contest than if Bush is up by double digits.
But presidential polling this far out from Election Day is only half science and half art -- exactly who is going to vote when November rolls around? -- and that's reason enough to regard an individual survey result with great caution. Sometimes, you get a better picture of the forest by looking at a lot of trees rather than just one. That's the case especially if one or two of the trees (polls) are just plain wrong.