Closed kuriwaki closed 2 weeks ago
I noted in the paper that Santa Cruz and Cumberland (https://github.com/kuriwaki/cvr_harvard-mit_scripts/issues/154#issuecomment-2318008762) are the only two counties currently in our release that do not record undervotes.
In this county of 20,000 voters (https://votedatabase.com/cvr/arizona/Santa%20Cruz/), the raw data does not appear to contain any indication of undervotes. Thus in our final data, there are no undervotes; this is misleading when we do undervote analysis with it.
For President and Senate we could fill in undervotes by seeing which cells are blank, but for district-level and non-federal elections this imputation becomes hard to do.
Good to know which machine vendors suffer from this flaw.