This isn't 32000 in 1 wave, though. This is ~2500 a year over 13 years. Even the answers given at the beginning of the study could have changed wildly if the same people had been polled at the end. And even if not, 4 people per city is not representative of an entire city at any given moment of time.

What demographics in China did they poll each year? Did they poll people of different racial profiles? Did they poll uyhgurs? Were the candidates selected randomly or were the assigned by the government? If the latter, were they coached or paid? Any number of things could throw off that study.