Reply #20: Ogegon might have the solution.

[jkd]

They could use a simple random sample of all voters with no worry about precincts, clustering, etc. Anonymity will be compromised, however. Some voters might object because the interviewer will know their personal information.

Oregon's system might be susceptible to the obvious type of fraud, but I believe wide-spread vote buying would be difficult to conceal.

At least when they match, we can have confidence that both the vote and the exit polls are legitimate.