> who really would want to risk real-world consequences for a random survey?
Why not? We don't know the baseline behavior and expectations of the respondents.
I think we differ on where our subjective Bayesian priors are situated.
I'm more at the 50/50 mark in that I think the results could be biased but at the same time, there's no necessity that they must be. Whereas most people on this thread automatically assume a priori that the survey results surely are biased, because China.
Why not? We don't know the baseline behavior and expectations of the respondents.
I think we differ on where our subjective Bayesian priors are situated.
I'm more at the 50/50 mark in that I think the results could be biased but at the same time, there's no necessity that they must be. Whereas most people on this thread automatically assume a priori that the survey results surely are biased, because China.