« The State of the Race With 9 Days to Go | Main | Time Vs. Newsweek »

Tracking the Tracking Polls

Here are today's tracking poll results (all data based on LVs, Rasmussen data include leaners):

Rasmussen: tie (49-49) from +3 Bush (50-47) 2 days ago
WP/ABC: +1 Bush (49-48) from +4 Bush (50-46) 2 days ago
Zogby: +2 Bush (48-46) from +2 Bush (47-45) 2 days ago
TIPP: +4 Bush (47-43) from tie (45-45) 2 days ago

So Bush's position is either weakening, strengthening or staying about the same. That certainly helps clear things up.

One is tempted to say movement in these polls can't possibly be providing much meaningful information on the state of the race.

Alan Abramowitz expands on this idea in his characteristically lucid fashion below:

For many political junkies, including myself, following the presidential tracking polls has become a daily obsession. We wait with bated breath each morning for Zogby to release his latest results. At the stroke of noon, we log onto the Rasmussen website to get our second daily fix. Finally, at 5 p.m. we eagerly await the latest update on the Washington Post/ABC News tracking poll. Some of us have even discovered a fourth tracking poll, done by an organization called TIPP (the TechnoMetrica Institute of Policy and Politics). The TIPP tracking poll usually releases its daily update sometime in the afternoon.

Tracking polls are different from other political polls. Most polls are done over several days to allow time for multiple attempts to reach those who do not answer their phone the first time. In contrast, in the case of tracking polls, all interviews must be completed the same day so callback attempts are limited or nonexistent. These daily samples are combined over three or four days, with the most recent dayís interviews added to the sample and one earlier dayís interviews dropped from the sample. The result is a kind of ďrolling sampleĒ that, theoretically, tracks day-to-day trends in support for the candidates.

Political campaigns have long used tracking polls to gauge voter response to the campaign and formulate strategy. In recent years, however, a number of media outlets have also been conducting tracking polls and reporting their results to the general public. Four years ago, for example, the Gallup organization conducted a tracking poll during the final month of the campaign. However, the results were so controversial that Gallup dropped its tracking poll this year. The problem with the Gallup tracking poll was that its results gyrated wildly from week to week, and sometimes even from day to day.

In order to avoid the kinds of problems that affected the Gallup tracking poll, the four tracking polls being conducted this year all weight their nightly samples based on certain assumptions about the demographic and partisan composition of the electorate. The result is that this yearís tracking polls have been much more stable than Gallupís 2000 tracking poll. And all of the tracking polls have produced similar results. On average, during the month of October, President Bush has had a lead of 1 percent in the Zogby tracking poll, 3 percent in the TIPP tracking poll, 2 percent in the Rasmussen tracking poll, and 3 percent in the Washington Post/ABC News tracking poll. Since October 12th, when TIPP joined the other three, the daily average of the four tracking polls has ranged from a 1 point lead for Bush to a 3 point lead for Bush with no evident trend.

So does it make sense to monitor the daily movements of these tracking polls? The answer is that if youíre hoping to learn something about real trends in support for the presidential candidates, it probably doesnít make sense. Thatís because there is no correlation between the day-to-day movements of the four tracking polls. In other words, they donít move togetheróeach pollís movements are unrelated to all of the other pollsí movements. For example, the average correlation between the daily movements of the Zogby Poll and the daily movements of the other three polls is -.18. The average correlation for the Rasmussen Poll is -.07, the average correlation for the TIPP Poll is -.09, and the average correlation for the Washington Post/ABC News Poll is -.12. The combined average for all four tracking polls is -.11. These weak negative correlations mean that there is actually a slight tendency for the polls to move in opposite directions.

What these results indicate is that the day-to-day movements of the tracking polls are essentially random. Rather than reflecting real shifts in voter preferences, the day-to-day movements of the tracking polls are simply reflecting sampling error. This doesnít mean that the overall results of these polls are wrong. In fact, the average margin between George Bush and John Kerry in the tracking polls has been very close to the average margin in other recent national polls. It just means that the day-to-day shifts in the tracking polls are probably not real and that the real level of support for George Bush and John Kerry within the electorate has not changed over the past few weeks: the presidential race has been very close since the beginning of October and it is likely to remain that way until Election Day.

So relax political junkies. Stop obsessing over the daily movements of the tracking polls and get a life! Follow the World Series. Follow your favorite college or professional football team. Follow the weather report. Follow something that is more real than the day-to-day movements of the tracking polls.

Words of wisdom from the good Professor. Heed them well and you'll get through the rest of the campaign with a considerably lower stress level.


Prof. Abramowitz says that there's no meaningful correlation among tracking polls on a day-to-day basis. I wonder if some correlation might be discovered if they're assumed to have different amounts of built-in delay. For example, Zogby might have a positive correlation with the numbers of the other polls from the previous day.

On the other hand, it might be equally useful to obtain a tiny bit of radioactive material and a neutron detector, and try to find political trends in the times between detections. Chicken entrails are good, too.

just curious. Has any organization run two simultaneous tracking polls at the same time - i.e. call 1000 independent people for one poll and another unrelated 1000 for a second, while keeping everying else - population selection, questions etc. the same

It would be interesting to see how much they varied.

Much has been written about the impact of cellular phones on the inaccuracy of polling.

What about Caller ID? I read somewhere that Democrats are more likely to screen their calls than Republicans are, thus less likely to pick up a call from someone who masks their number (or whom they don't know).

Any thoughts?

So what different between internal campaign tracking polls and public horse-race tracking polls? Views of the candidates on various aspects, opportunity & vulnerability, reaction to exterenal events &/or each campaign's efforts...? Is it like coffee bean traderes making a harsh overpowering brew to bring out less obvious traits -- campaigns aren't looking at horse-race numbers but the tactically useful info? They figure horse-race numbers can be deriveed as needed? Just a bunch of guesses.

The problem, as both Jimmy Breslin (http://www.newsday.com/news/columnists/ny-nybres214013687oct21,0,6406517.column) and Robert Cringley (http://www.pbs.org/cringely/pulpit/pulpit20041021.html) reported last week, is that none of these polls track people with just cell phones.

Bob G

I was wondering myself what these internal polls are. What I figure is that they have better voter registration data so they can normalize the polls better. Besides that you got me.

I can't find it at the moment, but earlier today I read on the washington post that Bush's internal polls show him doing worse than most of the polls in the media, both nationally and in the swing states! It said they have no leads outside the margin or error. That's great news to me, whatever the internal polls mean.


It's interesting that Zogby shows Bush widening his small lead (now just barely outside the margin of error) while the WaPo tracking poll shows the race narrowing (49-48).

How well did all these polls do in the last election? What is their track record? My understanding of what I've been able to find is that most had Bush with a lead until very late, when something showed up in the Zogby poll.

I agree with the Professor. Enjoy the World Series and see who wins Nov. 2 (or 3, or 4, or 5 or ?)

thanks. I now can relax. I am glued to the computer reading the latest tracking polls.

Nothing is going to lower my stress level before Tuesday; however, inadvertently, I clicked on a 2000 poll site, one week before that election. I initially thought I was viewing the 2004 polls for this week. All showed Bush leading except ABC/Washington Post. I cannot help concluding that most polls are conservatively biased.