About That Rasmussen Poll That Got Scott Adams in Trouble

As I’m sure everybody knows by now, Scott Adams of Dilbert fame got himself uber-cancelled by trying to draw the logical implications from a Rasmussen poll. The poll surveyed 1000 people, 130 of them black, and asked if they would agree with the statement, “It’s OK to be white.” Only 53% of the black respondents agreed.

Adams said that if 47% of black people could not agree with that statement, it was time to throw in the towel on race relations and to just find a safe place to live in an area where half the people don’t disapprove of your existence.

While Adams’ comments have been caught up in the typical Tweetornado that distorts and amplifies what he said, several people have criticized his reliance on the poll, claiming that the sample size wasn’t large enough to support Adams’ conclusion. Here’s one such critique from Red State.

But none of the critics I’ve read have actually looked at the math to determine how errant Adams may have been. So here’s my take:

Typical national-level poll survey 400 or 1000 people. 400 people gives you a 95% confidence that the poll results will only have 5% error. 1000 gives the same confidence for a 3% error. So the overall sample size used by Rasmussen was typical and adequate.

But when you start looking a sub-populations, you need to recalculate the required sample size. While the results for the general population meet the 95%/3% level, those values don’t pertain to the black subset of the population. For the ~40 million black population, you still need 400 respondents to get to that 95%/5% level. And for a volatile question of that nature, 1000 respondents would be better.

But how misleading was the poll information? 130 responses means that the error is 8.6% for a 95% confidence interval (I used this handy calculator to get that value). That says that there’s a 95% probability that the percentage of blacks who did not agree with that statement is between 38 and 56.

That’s still a problematically large number, and should motivate pollsters to do more focused follow-up studies.

In the meantime, I’ve read a lot of the criticisms of Adams’ comments, and they generally fall in to the obvious “he’s a racist” camp, or the “don’t believe the poll” camp.

But I haven’t seen any that say that the vast majority of the black population believe that it’s OK to be white.

Advertisement