Last year I testified before the NSW state Joint Standing Committee on Electoral Matters (JSCEM) about the issue of random sampling in NSW local government elections. You can read my submission, the transcript of the hearing, and the committee’s report.
TLDR: The random sampling system used to conduct NSW council elections is broken. The state government has accepted some great recommendations from the state JSCEM to fix this problem, and make some other improvements to instil more faith in our electronic counting systems, at least in NSW.
When a candidate is elected with a surplus in a single transferable vote election (such as elections to the Senate, all mainland upper houses, the ACT assembly and the Tasmanian lower house, as well as many councils), a proportion of their votes should flow on as preferences.
For most Australian elections, this is done through the use of a transfer value. All of the votes are passed on, but at a fractional value. So if you have 100,000 votes and the quota is 80,000, all 100,000 votes are passed on at 0.2 of a vote each.
This works pretty well, but was much harder to do back when counting was done by hand.
So in some systems, we used to only pass on some of the votes at their full value, and these ballots were selected randomly. So in the above example, 80,000 of those votes would stay with the elected candidate and the remaining 20,000 would be passed on at full value – even though those voters have already been able to have their say.
This system has a number of problems. It’s very difficult to randomise using a computer system (it’s impossible to do it perfectly) and it creates significant complications in making the computer code to run the count. It also is not replicable. A second recount could produce a different outcome.
This random sampling system is still used for the NSW Legislative Council (which would require a constitutional referendum to change) and for NSW local government (which could be changed with simple legislation).
The problems are worst for local councils with smaller numbers of votes being cast, larger fields of candidates and votes less likely to follow organised party tickets. There are numerous examples of council elections where the most likely outcome did not happen thanks to a biased sample, and a handful of cases where the result of an election was changed in a recount despite the margin of victory being larger than the number of errors found in the count.
The committee was surprisingly consensual, with all parties supporting a change in the counting system. I’m pleased to report today that the state government has responded to the committee’s recommendations and has accepted the recommendation. Hopefully this will mean legislation to implement this change.
The committee also made a number of other recommendations, all of which the government has supported. These include:
- Using the weighted inclusive Gregory method of calculating transfer values;
- Requiring council elections run by private providers to adhere to the same standards for scrutineering and provision of data as the NSW Electoral Commission;
- The NSWEC working to develop a policy to make it easier for scrutineers to examine paper ballots, electronic records and data entry records (specifically in the context of computerised counting making it harder to follow the count);
- Developing an audit process to check that paper ballots match the data entered into the computer, which could be observed by scrutineers;
- That source code for counting be subject to an external audit at least every five years (the NSWEC says this is consistent with their current practice); and
- That candidates can no longer insist on a recount if they pay for it (with the Commissioner still able to decide to hold a recount).
This was a very productive process and I hope we’ll see legislation where it is needed, and these changes in policy will be in place for the next council elections. A lot of these procedures could also apply to NSW state elections, so I’m hopeful this will have broader flow-on effects. Thanks to those on the committee who gave me the opportunity to testify.