Almost all of the drug users I know are unemployed social outcasts. One of them at least somewhat tries.
Conversely, all of the drug users I know are employed and earn between 1x and 3x the national average household income. At jobs where they're highly respected for their contributions.
Again, maybe that's just anecdote. But the "anecdote" criticism cuts both ways. The only real difference is that I'm not justifying jailing people on the basis of personal anecdote.
You're saying that by sampling opinions that are formed in small neighborhoods around voters, we can get a picture of ground truth -- even if reality is distorted in various ways around those small neighborhoods.
This can work, but not in the case where there's something masking a portion of the data set such that personal anecdote isn't indicative of underlying reality, even locally.
In that case, anecdotes aren't even locally accurate, and that effect amortizes over the voting population to result in a globally inaccurate picture of reality.
Which is the case here. Criminalizing and stigmatizing drugs means that grandma doesn't necessarily know successful son-in-law smokes pot on the weekend, for example.