On the reliability of Gallup

Two days ago, I reported that Gallup’s seven day tracking average showed Romney up 51 to 45 percent. Yesterday Romney’s Gallup lead went up to 52 to 45. Reader Andrew B. says that Gallup polls are deeply misleading and he confidently predicts an Obama victory in the election. Meanwhile Jennifer Rubin of the Washington Post, predicting a Romney victory, says that “No candidate over 50 percent in Gallup this late in the race has lost.” At first glance Rubin’s claim seems unbelievable—Gallup has never been wrong? What about 1948, when all the polling organizations including Gallup famously predicted a Dewey victory and never got over their embarrassment at being so wrong? Well, since there were two minor parties in that election, the Dixiecrats and the Progressives, which together won almost five percent of the national vote, we must assume that Gallup showed Dewey leading, but with less than 50 percent. (The final result was Truman 49.6 percent, Dewey 45.1 percent, Dixiecrat Strom Thurman 2.4 percent, and the Progressive Henry Wallace 2.4 percent.) So Rubin’s statement about Gallup’s reliability would seem to stand, or at least it’s not obviously wrong.

- end of initial entry -

Dave T. writes:

I thought Andrew B did a good job of explaining why Obama would win the election given everything we knew prior to the first debate.

LA replies:


James P. writes:

Ace of Spades HQ linked to this cartoon and noted, “it’s not hard to come up with plenty of ‘rules’ which never fail—well right up until they do.”

Paul K. writes:

The cartoon linked by James P. is brilliant. And—synchronistically!—it reminds me of the useless George Will, discussed in the next post. During the 1988 campaign, I recall Will constantly saying that it was unlikly that George H.W. Bush would be elected because no sitting vice president had been elected since Martin van Buren. You might wonder why I remembered such a trivial fact; I remember I thought that to predict the outcome of an election on such a meaningless precedent indicated that Will was a pedantic fool.

LA replies:

I agree the cartoon is very apropos. Perhaps the most important “No candidate with characteristic X has ever been elected” rule in this election is: “No Republican who has lost Ohio has ever been elected.” Now this is true, but misleading, since for a long period in American history, Ohio was an overwhelmingly Republican state. Between 1868 and 1923 many presidents were Republicans from Ohio. The Republican candidate almost always won Ohio. Secondly, there are plausible scenarios by which Romney could lose Ohio but still win the election. So I wouldn’t be surprised if this particular “has never won” is toppled next month.

I also read this morning, I think in John Podhoretz’s column in the NY Post, that the Obama campaign is banking everything on defeating Romney in Ohio. They are spending unprecedented amounts on negative ads in that state on the basis of the belief that if the Republican loses Ohio he must lose the election.

Posted by Lawrence Auster at October 19, 2012 11:48 AM | Send

Email entry

Email this entry to:

Your email address:

Message (optional):