> You can never repliacte a census using random surveys. It's the holy Grail of data. True. As a matter of fact it's good to use the census as a sort of period

> You can never repliacte a census using random surveys. It's the holy Grail of data. True. As a matter of fact it's good to use the census as a sort of periodic anchoring point for your surveys (which can happen more often). To completely give up on the closest thing to ground truth that can be had is foolish.