Biased, maybe. But the sample sizes are pretty big, whether "bigness" refers either to absolute numbers or to being big enough such that whatever they're estima

Biased, maybe. But the sample sizes are pretty big, whether "bigness" refers either to absolute numbers or to being big enough such that whatever they're estimating will come pretty close to the true value of that variable for similar populations. I mean, the smallest sample is 70k plus.