[tzs]

> They also only polled 463 people and are extrapolating the 1.8 million number based on the fact that 13% (~60 people) included "unemployment insurance" as one of the factors that influenced their decision.

463 people drawn from population of 14 million, assuming unbiased sampling, with a 13%/87% split on their answers to a yes/no question would have a confidence interval of about 3 at 95% confidence, so it is reasonable to conclude that the correct percentage is probably in the 10-16% range, or 1.4 million to 2.2 million.

At 99% confidence, the confidence interval would be about 4, giving 1.26 to 2.38 million.