Donate to NRO Today


NRO BLOG ROW | THE CAMPAIGN SPOT |  ARCHIVES    SEARCH    E-MAIL    RSS




Thursday, March 27, 2008


HORSERACE

Picking Over That Counterintuitive NBC/WSJ Poll

The Wall Street Journal poll is generating buzz. But I have a lingering skepticism about its counterintuitive results — that Hillary Clinton has seen her favorable numbers drop in the wake of the Jeremiah Wright scandal more than Barack Obama.

NBC's Chuck Todd writes, "In addition, we oversampled African-Americans in order to get a more reliable cross-tab on many of the questions we asked in this poll regarding Sen. Barack Obama's speech on race and overall response to last week's Rev. Jeremiah Wright dustup."

But is the oversample included in the sample as a whole? In response to the question of what race you are, the polls says 11 percent said "African American." But the sample size for African-Americans was 177. They had 700 initial respondents, then another 100 to enlarge the African-American sample. That puts them at either 22 percent of the sample (177 out of 800) or 25.2 percent of the sample (177 out of 700), which either way is larger than the usual 13 percent of the general population, or 11 percent of the 2004 electorate (according to exit polls).

If your sample doubles the percentage of African-Americans who usually show up on Election Day, wouldn't we expect Barack Obama, the candidate who is winning an overwhelming share of the African-American vote, to do well compared to other polls? And wouldn't this hide any Wright fallout?

If Hart/McInturff say that 11 percent of their respondent pool was African-American — and the extra were only used for questions specifically aimed at measuring African-American opinion — then all is well, more or less. But note that even with the oversample, the margin of error on questions to that respondent pool is still +/- 7.4 percent, higher than most regular polls.

Beyond that, this poll is generally out of step with other results. (After Wright, Obama's favorables among white voters were essentially unchanged, and his unfavorable rating went up 2 percent? Really? And 16 percent of whites describe their feelings toward Jeremiah Wright himself, not Obama, as "neutral"? When asked about their feelings toward Jeremiah Wright, 49 percent of whites say they don't know the name or aren't sure?)

As Michael Barone notes, in Rasmussen, Obama's unfavorables jumped after the Wright and his favorables slid;  in the daily tracking poll, McCain has beaten Obama by at least six points since March 17 and leads by 10 yesterday. SurveyUSA shows Obama losing Ohio by 7 to McCain, losing Kentucky by 36, Missouri by 14, Minnesota by 1, a tie in Massachusetts. With one exception, the Pennsylvania numbers look terrible for Obama. 




 





 

© National Review Online 2009. All Rights Reserved.

Home | Search | NR / Digital | Donate | Media Kit | Contact Us