A discrepancy of 200 votes to 203 votes is 1.5%, which gets caught in our 1% criteria. However, these are quite small vote discrepancies, and often votes for small parties.
We could make an exception to the 1% rule if, say, the absolute difference is only 3 votes or less. That would release 8 more counties (see the max_diff column here) I think:
read_excel("compare.xlsx", sheet = 3) |>
filter(release == 0, state != "DISTRICT OF COLUMBIA") |>
mutate(diff = abs(votes_v - votes_c),
diff_pct = abs(votes_v - votes_c) / votes_v) |>
mutate(n_voters = sum(votes_v[office == "US PRESIDENT"], na.rm = TRUE), .by = c(state, county_name)) |>
# discrepancies that are flagged
filter(diff_pct >= 0.01) |>
# largest discrepancy
summarize(
max_diff = max(diff),
max_disc = max(diff_pct),
.by = c(state, county_name, n_voters)) |>
# sort from small to large
arrange(max_diff) |>
print(n = 20)
A discrepancy of 200 votes to 203 votes is 1.5%, which gets caught in our 1% criteria. However, these are quite small vote discrepancies, and often votes for small parties.
We could make an exception to the 1% rule if, say, the absolute difference is only 3 votes or less. That would release 8 more counties (see the max_diff column here) I think: