Reply #16: FWIW [View All]

[Febble]

I did a chi square on those numbers. First batch and third are not significantly different from each other. Second batch is significantly different from first and third.

But all that means is that the difference between the second batch and the other two was not simply due to sampling error. In other words it means that the 100 precincts were "drawn from a different population" than the last 91 or the absentees.

Nothing intrinsically improbable about that - it just means that the second batch was not a random sample from the whole. The question you could ask is: what characterised the demographics of the second and third batch? And why did the third batch resemble the absentees?

But it is possibly worth noting that it is the batch with the smallest margin that is out of line with the other two, not the batch with the largest margin.