I’m as fascinated by the headline, “Blacks Shift To Obama, Poll Finds” as the next guy.
But the poll has the detail:
“This Washington Post-ABC News poll was conducted by telephone February 22-25, 2007, among a random national sample of 1082 adults, including an oversample of 86 black respondents.”
Hmm. As of July 1, 2005, the estimated population of black residents in the United States, including those of more than one race, made up 13.4 percent of the total U.S. population. By my back of the envelope calculation, if the sample of 1082 adults perfectly matched the U.S. population proportions by race, 145 respondents would be African-American.
Just 145? (Remember, the 86 oversamples – additional calls made to reach a particular demographic - are included in the national sample.) Is that the number of respondents who prompted this analysis:
Clinton’s and Obama’s support among white voters changed little since December, but the shifts among black Democrats were dramatic. In December and January Post-ABC News polls, Clinton led Obama among African Americans by 60 percent to 20 percent. In the new poll, Obama held a narrow advantage among blacks, 44 percent to 33 percent. The shift came despite four in five blacks having a favorable impression of the New York senator.
A sample of 145 (or even a little more) isn’t a small one, but it’s not exactly huge, either. A sample size of 200 results in a margin of error of 7 percent (at a 95 percent level of confidence); a sample size of 100 results in a margin of error of 10 percent.
From last month, Obama is up 24 percent, Hillary is down 23 percent. Any way you slice it, that’s a significant shift. But one can’t help but wonder if 145 respondents is enough to justify a page A1 headline, “Blacks Shift To Obama, Poll Finds.”