realjohnboy wrote:But when the article states that 1200 folks were African-American and 1000 were white, does that not cause you to suspect that this survey has flaws as being representative of "random" Americans?
Not necessarily, no.
Pollsters sometimes "oversample" a minority group of respondents, because if they would merely sample them proportional to their share of the overall population, their number would be too small to be statistically reliable. This allows them to draw meaningful conclusions even about the opinions of this small minority group.
E.g., if you interview 1,000 randomly selected people, you'll get something like 100 African-Americans, and you basically wont be able to say anything meaningful about the opinions of blacks. So instead, you might survey 1,000 randomly selected people
and 1,000 randomly selected African-Americans, so you can make meaningful observations about both groups separately. Which is roughly what this poll did.
But of course, when the pollsters then write down the report about the opinion among the population as a whole, they dont just take the average of 1,000 whites/hispanics etc and 1,000 blacks and present the results as "what Americans think". Instead, they weigh. They adjust back.
They know that, say, 10% of Americans is African-American, so they make the results of those 1,000 survey forms weigh as 10% of the score, and the 1,000 survey forms from all other groups weigh as 90% of the score.
Thats apparently also what happened here:
Quote:More than three quarters, 76 percent, of respondents in a CNN/Essence Magazine/Opinion Research Corp. poll said the country is ready to be led by an African-American [..]
Of the white Americans surveyed, 78 percent said the country is ready, as opposed to 69 percent of African-Americans polled.
The wording ("76 percent of respondents") is a bit confusing here, because the 76% must be the result of weighing. See how the result among African-Americans is 69%, and among whites 78%, and the "total" result is 76%. They interviewed just as many whites as blacks, and yet the "total" result is much closer to the "white" number than to the "black" number. So they must have weighed the results from the two groups into the total on the basis of their population size.
This is fairly common. And with a total of over 2,000 respondents, this poll sounds very solid at least on that count - most election polls you see cited in the media are based on just 1,000 respondents, or less (sometimes just 500-600).
Finally, it's noteworthy that blacks are actually more pessimistic than whites, according to this poll, about whether the country is ready for a black president -- so the weighting to make the totals from the white respondents count for x times as much as those from the black respondents in the overall results had as effect that the percentage of those who said America was ready went
up!
Re the poll having been done by Essence, Fishin' said it already: that the poll was commissioned by Essence doesnt mean it was a poll of Essence readers. The poll was conducted by Opinion Research Corp, and they'll have done a random sample of Americans.
One thing you're totally right to be sceptical about though: "folks will always claim to not be prejudiced, but when they actually go into the voting booth...". Thats the major weakness of this kind of research. One way pollsters sometimes try to work around this is by asking people, not just whether they would vote for a black people or Mormon or woman or whether they think "America" is ready; they also ask whether they think any of their
neighbours would refuse to vote for etc. It's surprising how many people are quite able to acknowledge what their neighbours really think that they'd never admit about themselves...
