Earlier this morning we wrote about the obvious sampling bias in the latest ABC / Washington Post poll that showed a 12-point national advantage for Hillary. Like many of the recent polls from Reuters, ABC and The Washington Post, this latest poll included a 9-point sampling bias toward registered democrats.
"METHODOLOGY – This ABC News poll was conducted by landline and cellular telephone Oct. 20-22, 2016, in English and Spanish, among a random national sample of 874 likely voters. Results have a margin of sampling error of 3.5 points, including the design effect. Partisan divisions are 36-27-31 percent, Democrats - Republicans - Independents."
Of course, while democrats may enjoy a slight registration advantage of a couple of points, it is nowhere near the 9 points reflected in this latest poll.
Meanwhile, we also pointed out that with huge variances in preference across demographics one can easily "rig" a poll by over indexing to one group vs. another. As a quick example, the ABC / WaPo poll found that Hillary enjoys a 79-point advantage over Trump with black voters. Therefore, even a small "oversample" of black voters of 5% could swing the overall poll by 3 full points. Moreover, the pollsters don't provide data on the demographic mix of their polls which makes it impossible to "fact check" the bias...convenient.
More
Pfffft! Another worthless poll.
ReplyDelete