(Skip to widget.)
I had hoped on this page to talk about inferring rates of accidental informality in the House of Reps based on Senate preferences – my idea was that since House informal votes are relatively common in much of western Sydney, and '1 only' Senate votes are also relatively common in western Sydney, it would follow that the Senate informal rate would be quite low, and we could attribute most of the difference between HoR and Senate on voters not knowing that they needed to write all preferences for the House vote.
This idea doesn't work. It turns out that informal Senate votes are also relatively common in western Sydney, and that the sum of the Senate informal vote and the '1 only' Senate votes can be much greater than the House informal vote. Perhaps we can attribute the high '1 only' Senate voting rate in NSW to their Legislative Council voting.
So instead this page is a grab-bag of miscellaneous table and mapping options. There's the House informal vote; the Senate informal vote; the above-the-line Senate voting percentage (of formal votes); the percentage of ATL votes marked with a '1'; the percentage of ATL votes marked with a '1' that went on to give at least a second preference; the percentage of ATL votes marked with a tick or cross (under the Electoral Act, these are equivalent to a '1'); and the percentage of such votes that went on to give at least a second preference.
I've assumed that the AEC's data-enterers were consistent in coding ticks and crosses as '/' and '*' in the formal preferences file, but since they are functionally equivalent to '1', it's no great loss if this coding is inaccurate. Certainly there are non-trivial patterns in the ticks and crosses data – western Sydney (again) has more of these votes than other areas. But in any case we're talking about a small proportion of votes, with the highest electorate on this table being Blaxland, at 1.76% of ATL votes being marked with a tick or cross.
As shown in the table below, a greater proportion of votes with a tick have further preferences written, compared to votes with a cross. Perhaps some slanted '1's are confused with ticks; I don't know.
Plotting the proportions of ticks and crosses by electorate, it's clear that they're closely correlated (0.94 nationally; 0.8 excluding NSW).tick/cross scatter plot.
So despite the differing further-preferencing rates, I've grouped the ticks and crosses together in the table/map options below. A CSV file with raw vote tallies is here.