Stomping On The 2004 Exit Poll's Grave (And Some Other Myths)
Print Friendly and PDF

[The Sailer Election 2004 Series: Sailer Strategy Wins Another For GOP—But How Much Longer?; Bush Didn't Win 44% of Hispanic Vote —The Smoking Exit Poll; Another Nail In The Coffin Of Bush's "44% Hispanic Share";  I Told You So Department: Only Bush Boosters Now Believe 44% Hispanic Vote Myth]

It may not be as flashy as's rout of Dan Rather. But I really have to congratulate me (and VDARE.COM) for routing the exit poll-fuelled media myth that George Bush made a big breakthrough among Hispanics this year.

The internet rules!

Edison-Mitofsky, the firm that conducted the troubled 2004 National Exit Poll (NEP), has now issued a long report (PDF) reviewing its own performance. It offers some important nuggets about what really happened last November.

  • E-M's analysis of the exaggerated Bush share of the Hispanic vote (pp. 59-62) confirms my diagnosis of what went wrong, as I elaborated in, see above).

As I'd discerned, Bush did better among Hispanics on the long form questionnaire that Edison-Mitosfsky had given out at 250 polling stations (total sample size of 12,219) than on the short form questionnaire distributed at 1,469 locations (sample size of 75,537).

The long form exaggerated the national and regional Bush share of the Hispanic vote—especially the bizarrely high figure in the South region, where Bush supposedly won 64% of the Hispanic vote, even though he carried only 56 percent in Florida and 49 percent in Texas. (Which was reduced from the initial announcement of 59 percent).

Nationally, Bush supposedly lost among Hispanics only by 53-44 on the long questionnaire, but got whipped 58-40 on the larger sample size short form.

Back on November 7, I wrote: "The big difficulty with an exit poll is coming up with a representative sample of polling places. Apparently, the NEP failed to do this." That's exactly what went wrong with the National/Regional exit poll's Hispanic share, as Edison-Mitofsky now admit.

  • The Edison-Mitofsky report also contains an interesting table (p. 59) showing six more demographic groups where the widely publicized National figure for Bush's share disagreed substantially with the sum of the State exit polls.

Here's Bush's share for each:

  National States
  (Small Sample) (Large Sample)
Hispanic 44% 40%
Asian 44% 39%
Age 75+ 45% 48%
Jewish 25% 22%
Mormon 80% 76%
Muslim 6% 13%
Income >$200,000 63% 60%

All of these are small and geographically-clustered groups. So the sum of the State exit polls is inherently more trustworthy and than the smaller sample size National poll.

My comments:

  • The Asian mirage. The news, reduced Asian share is worth noting in the context of the President's plan to increase immigration. Here's a largely prosperous, law-abiding, and socially conservative "model minority." Yet Asian-Americans apparently can't stand Mr. Bush. They gave him only 39 percent of their votes, compared to 58 percent among non-Hispanic whites.

  • The Neoconservative Mirage. Bush's 22 percent share of the Jewish vote, although reduced from the small sample estimate, is of course slightly better than the 19 percent he achieved in 2000. But then, John Edwards had replaced Joe Lieberman as the Democrat's VP nominee. So you'd expect a Republican to win back some conservative and moderate Jews who liked Lieberman. Compared to how well Republicans did from 1976 through 1988, when their share of the Jewish vote ranged from 31 percent to 39 percent, 22 percent is very bad.

And, when you consider how much of the neoconservative invade-the-world-invite-the world foreign and immigration policies Bush adopted as his own—well, 22 percent is unbelievably awful.

What this shows is that neoconservatives can make a big noise, but they can't deliver the vote. With Jews casting only 3 percent of all votes, the neoconservative vote comes out to only 2/3rds of one percent of the electorate.

To put in Texan terms the President ought to understand, the neoconservatives are all hat and no cattle.

  • The Muslim Mirage: It's not surprising that there's a big difference between the small sample and large sample figures for Bush's share of the Muslim vote (6 percent vs. 13 percent), because the total quite tiny—only 1 percent (compared to 3 percent for Jews). And of course, that's rounded. It would be useful to learn whether the unrounded Muslim proportion of the total vote was actually closer to 0.5 percent or 1.4 percent—in the 2002 election, it was only 0.3 percent.

Either way, it's not worth Grover Norquist's time.

[Steve Sailer [email him] is founder of the Human Biodiversity Institute and movie critic for The American Conservative. His website features his daily blog.]

Print Friendly and PDF