Print Columns   |   Web Chats   |   Blog Archives   |  

When Is A Poll Not A Poll? (Virginia Senate Edition)

The big news organizations haven't done much of their expensive polling yet on Virginia's Democratic primary in the U.S. Senate race for the seat now held by George Allen. And chances are that most, if not all, news companies will decide to pass on doing any polling ahead of the June vote, saving their money for the fall general elections.

But this volatile news environment we live in abhors a vacuum, so here come the bloggers. Virginia's Not Larry Sabato, a political blog that used to be an anonymous venture but is now widely known to be written by campaign worker Ben Tribbett, has taken its own poll--using a controversial but cheap methodology that, if it were to become widely accepted, could alter the dynamics of state political campaigns.

Virginia's political bloggers are in a bit of a tizzy over Tribbett's polling method, which involves using robot-phone calling machines to dial up hundreds of Virginians in search of a random sample. The machines can call folks at home at virtually no cost, but polling professionals question just how accurate such a poll can be when we know that a great many people hang up as soon as they hear that it is a machine that is disturbing their peace (count me as one of those folks.) Still, some professionals concede that some robocalling polls have proven to be reasonably accurate, including some by the Survey USA firm.

What Tribbett found is a strong lead for Jim Webb, the former Republican who was Secretary of the Navy under Ronald Reagan, over northern Virginia telecommunications executive Harris Miller, but even if you buy the poll's methodology, there's a huge caveat: Well more than half of those polled--57 percent--are undecided, so what's being measured here, at best, is the leanings of the politically involved.

Which, of course, is who's out there reading the political blogs. (By the way, if you're looking for the real Larry Sabato's take on the Virginia Senate race, here's the University of Virginia political science maven's latest report.)

And if you'd like to see another use of robocalling in a Virginia poll, here's the Rasmussen poll from last week, which shows Miller matching up against Allen slightly better than Webb does, but with the incumbent handily defeating either challenger. Of course, it's very early, and the poll does show some deterioration in Allen's numbers over time. Rasmussen, like Tribbett's maiden effort, is considered not-airworthy by the broadcast networks because of the robo-calling. Similarly, the Post and other major newspapers do not report the results of such polls (except when they slip in like this under the guise of talking about polling methodology. We live in a slippery time.)

Whatever happens in the wonderful world of polling methodologies, Tribbett's experiment has all sorts of delicious potential. In many local races, candidates who cannot afford to commission their own surveys operate more or less in the dark, going on good old gut instincts as they try to figure out where to aim their efforts. If any random blogger can rustle up a poll by befriending an old pal at a robocalling marketing company, it could become easier for outsider and unknown candidates to figure out niches of voters that could be worked hard for relatively little money. Politics could get a bit more interesting: Oddly enough, adding that little bit of extra predictive power to candidates' arsenals could make elections a tad more unpredictable, and that is something this country desperately needs. Of course, an epidemic of polls with lousy, unreliable methodologies could also poison the waters for legitimate polls--and there are those who would also welcome that development.

One more bit from Tribbett's poll: He asked Democrats which were the most important factors in their decision about whom to support against George Allen. The top answer was "ability to beat Allen," followed by the Iraq war. No other issue came close to those two.

By Marc Fisher |  April 26, 2006; 7:30 AM ET
Previous: Washington's Polite Quintet | Next: Virginia's New Odd Couple


Please email us to report offensive comments.

Seems to me the "roto" dialers are a violation of the do not call regulation. If not, they ought to be.

Posted by: Charley Brown | April 26, 2006 10:11 AM

Robo calls by political or survey research firms are not a violation of the do not call lists and are protected by law the 1st amendment.

Posted by: Gonzo | April 26, 2006 10:29 AM

forgot an "and" at the end. sorry.

Posted by: gonzo | April 26, 2006 10:30 AM


You fail to mention a couple of key points that a third year journalist should not miss.

1. Ben Tribbett has "endorsed" Webb, and has been active in smearing Miller. Tasteless. Absolutely tasteless.

2. Ben Tribbett is good friends with Leslie and Larry Byrne. Both of them cannot stand Miller, and Larry has made this campaign personal. It is well known in FCDC circles that Larry and the Webb staff have been passing Miller-smear information to bloggers, including Tribbett. Call Ginny Peters.

3. Tribbett has not released all of his data and methodology on his website, at least not at the same time. This is wrong, and casts doubt on the pollster's integrity. The whole poll comes off as phony.

4. If bloggers want to be taken seriously when also acting as pollsters, they must remain neutral, provide all information at the same time (not string readers), and provide a reasonable margin of error that lines up with other credible polls. +50% undecided means that Tribbetts poll is not very credible. We don't even know which phone list he used, it was only conducted in Northern Virginia (a big no-no when a race is state-wide!), etc.

Your blog makes it seem like you've been schmoozing a little too much with Tribbett.

Posted by: Fairfax, VA | April 26, 2006 10:54 AM

Why the hangup on methodology? I thought random-digit dialing was pretty standard for social surveys and political polling different? It's hard to get excited about this methodology if you don't explain the standard way of doing it.

Posted by: h3 | April 26, 2006 2:35 PM

RDD still needs to be balanced at the end based on the expected electorate. If you weight a sub-group wrong, you can skew the results.

Posted by: RDD | April 26, 2006 4:37 PM

And what's up with polling only N. VA for a U.S. Senate race? Useless information.

Posted by: Burke | April 26, 2006 4:40 PM

Oh, ok - so it's not random-digit dialing that's the problem, it's people who aren't pros doing political polling? That makes more sense. I can see how you'd want someone with actual knowledge of statistics and demographics and whatnot to be doing your polling.

Posted by: h3 | April 27, 2006 11:16 AM

hello! online directory. About DIRare, Search in Business Category, Yellowpages search. From online directory .

Posted by: online directory main | April 27, 2006 1:28 PM

Perhaps I'll start spamming people with my own survey and call it a "poll." Sure, I'm no statistician, I'm not even very good with math. Heck, I live in my parents' basement. But I'm running with the big boys now!

Posted by: Don Hogan | April 29, 2006 12:04 PM

Wait one minute.

Are you implying that Ben Tribbett lives in his parent's basement?

That is cruel. The guy is probably a struggling college student at GMU or NVCC.
Give the guy a break! Democrats need aspiring, young, optimistic, future political leaders like Ben!!

Posted by: Fairfax, VA | April 29, 2006 12:59 PM

The comments to this entry are closed.


© 2010 The Washington Post Company