Sometimes They Don't Need No Steenking Masks

The Washington post recently took a poll, described in the last paragraph of the article as:

The Post-ABC poll was conducted Jan. 20-23 among a random national sample of 1,003 adults, including interviews on land lines and with cellphone-only respondents. The overall margin of sampling error is plus or minus 3.5 percentage points.

There is not enough information -- by far -- to evaluate the adequacy of this procedure. (What sampling frame did they use to access the adults? Land-lines tend to access whole households; cellphone-only will pick up a lot of teens. Was this a two-stage cluster sample? The only thing we know for certain is that the sampling error was probably larger than 3.5%-points. Why? Because the results of any poll are usually less certain than we think. As near as TOF can figure that is a sampling error based on a simple (unstratified) random sample, a very difficult thing to obtain.

But the scariest result of the poll is not
Read more »