At the time of the July 2016 federal election, around 15.7 million names were on the electoral roll, some 14.3 million of whom turned out to vote. They either voted in person — on election day or beforehand — or by post, and between them they cast 13.5 million formal lower house votes.
In the weeks following the poll, the Australian Electoral Commission scanned all the paper printouts of the roll and found 18,343 people whose names had been crossed off more than once. In September that year they sent a letter to each of these apparent multiple voters.
The AEC does this after every election. In 2013 the number was a little higher, at 18,770.
Is 18,343 “multiple voters” a lot? How many of them were fraudulent? Should it worry us? How do we stop it happening?
The 18,000-plus figure averages out to about 120 per House of Representatives electorate, which could obviously make a difference in a tight contest. The closest seat in 2016 was Herbert in Queensland, which Labor snatched from the government by thirty-seven votes. Not surprisingly, Coalition members of the parliamentary electoral matters committee have taken a particular interest in that result.
So has the AEC, which asked the Australian Federal Police if they wouldn’t mind having a close look at forty-two Herbert voters who appeared to have voted more than once.
The wallopers investigated and concluded:
Thirty-three (78.5 per cent) were attributable to potential AEC error (electoral roll marking error, incorrect advice re: absentee voting, failure to identify/destroy duplicate records);
Five (12%) were attributable to potential voter error (genuine confusion with process, mental illness or medical issue, language issue); and
Four (9.5%) were attributable to there being no obvious explanation (no evidence of intent to vote more than once).
Regarding the country as a whole, the AEC found that “among the number of electors who provided an admission to multiple voting, 84 per cent of cases were the result of age, confusion, or poor comprehension.”
Words like “admission,” “evidence” and “intent” make it all quite vague. And if a voter whose name was marked off twice simply swears that they only voted once, there’s little the AEC can do.
Those 18,343 people equal 0.13 per cent of the electoral roll. Again — is that a lot or a little? The vast majority turn out to almost certainly be mistakes made by AEC staff. Of the rest, we would expect a tiny proportion of voters to turn up mad, bad, drunk, senile, mentally ill and/or combinations of the above. Humans do all sorts of strange things.
But it’s also reasonable to believe that a tiny number, around a tenth of all multiple mark-offs, deliberately voted more than once — as a lark, to prove some point or to get their preferred candidate elected. (Some fifty-nine names were crossed off three times or more, the largest number of cross-offs was sixteen; there was definitely mischief-making in most of them.)
At a general election, any coordinated multiple-voting effort would need to be organised across a great number of electorates, because we never know in advance which seats will be close enough for it to matter. Very few people had Herbert on the radar as a possible Labor gain in 2016, for example.
But something like last year’s Wentworth by-election, with the numbers so close on the floor of the House, would be enticing to fraudsters. (As far as I know the AEC hasn’t released multiple-voting data for Wentworth, and no suggestion of fraud has been made there. The final margin was a healthy 1850 votes.)
What’s important to remember is that elections need to be not only fair but seen to be fair. That 18,000-plus number is a gift to those sections of the media and social media that specialise in sowing distrust, and it’s not hard to imagine many people feeling troubled by the ease with which voter impersonation is theoretically possible.
So, what further safeguards can be put in place to minimise these multiple cross-offs?
The most commonly suggested solution is mandatory voter ID. In this, as in so much of our political discourse, we are influenced by the longstanding and toxic American debate, where ID laws are used by parties to try to exclude certain categories of voter. But many countries use voter ID without partisan considerations, though a lot of them (particularly in Europe) have national ID cards that all citizens must carry.
Voter ID makes a lot of sense but has severe drawbacks. Under a photo ID system a huge majority of Australians would find voting as easy as they do today. But a sizeable minority — perhaps 10 to 15 per cent who don’t carry a driver’s licence or proof of age card around with them — could run into problems.
In 2014 the Queensland LNP government introduced an ID system that was at the benign end of the spectrum. The Electoral Commission of Queensland would send out a letter that each voter could then use as identification. In addition, a range of possible ID documents, from an electricity bill (without a photo) through to a passport, could be used. And someone without any ID at the polling station could fill out a declaration vote — accompanied by a signed statement in an envelope — and if the ECQ found no problems then the vote would be included in the count.
Not surprisingly, increased delays were reported at the following election. And that system would not, for example, stop me overhearing you in the pub say you’re not going to vote, and voting in your stead. But having to sign a bit of paper would discourage that. ID can be forged, commission letters stolen from letterboxes, but the signature discourages deliberate mischief-making. On the other hand, staff will still make mistakes.
In 2015 Queensland’s Labor government abolished voter ID. Unfortunately, the ECQ doesn’t seem to have conducted useful research into the former system’s effectiveness.
Another option is “real-time electronic mark-off,” which both the electoral matters committee and the AEC are in favour of, and which has been tried out at recent federal by-elections. Instead of those big books, the electoral roll — connected to the central database —is held, live, in the hands of staff. This ensures that names aren’t marked off twice.
In New South Wales the state electoral commission is trying these out for early voting this year, and early last week internet connection problems caused them to turn people away. The most recent ACT and Northern Territory elections used real-time electronic mark-off virtually everywhere, and the AEC claims great success in all but eradicating multiple voting in those two jurisdictions.
But we only have to briefly contemplate how this works to realise that much of the improvement is illusory.
Say you rock up on election day to vote (either in the context of voter ID or not) and are informed your name has already been crossed off. Either someone has impersonated you or a polling official has made a mistake. “Not to worry,” you are assured, “fill in this declaration vote.” This is time-consuming, tedious and, most importantly (though presumably you will not be told this), almost certainly a waste of time; because your name has already been marked off, your vote won’t be counted.
In this way multiple voting is “eliminated.” But for real-time mark-off to achieve its purpose, you have to assume legitimate voters attend first and impersonators dawdle along later. And any accidental marking-off of the wrong person would, instead of creating an apparent double vote, simply disenfranchise the voter.
As New South Wales showed, technical issues are also a factor. And a live roll on election day? The potential for malicious hacking activity ramps up the blood pressure (in this writer anyway).
(Note the difference between this live roll and having standalone copies of the entire roll in every polling official’s hand. The latter is a very worthwhile aim for the AEC, with little downside except the huge cost.)
It’s possible that accidental double-marking is less likely on electronic than on paper rolls. Or it might be more prevalent.
In a joint submission to the electoral matters committee, the AEC and the Federal Police suggested CCTV cameras at polling stations. Effective? Yes, but also uber-expensive, and they might discourage some people from voting at all. It’s very Big Brother, and not a goer in the short to medium term.
Given that the vast majority of the 18,000-plus were staff errors, and the exercise is largely concerned with fortifying public confidence, does it make sense to direct resources towards reducing that figure to, say, 5000 or so? Having two workers instead of one, to double-check that the correct name is crossed off, would be horrendously costly.
A less alarming variation than CCTVs might be sound recorders at each table, so that staff can go back and listen to the name that was uttered at the relevant time. But this would only work in conjunction with time-stamped electronic mark-off.
It’s likely that one day we’ll nearly all be voting online with our MyGov accounts, with the few lagging technophobes doing it in person, verified by biometrics, facial recognition or computer chips.
In the meantime, a mild ID system, such as Queensland’s former arrangements, in the context of a non-live electronic roll across the country, might be worth considering. (It’s obviously too late for this year’s election.) Perhaps also those audio recordings. Extra resources would be needed, especially for the electronic roll.
It would be crucial for governments to fund any of these solutions adequately to avoid increased delays at polling stations. It can’t be done on the cheap.
Voter ID might be a solution in search of a problem, but perceptions are important. In our social media–driven, tinfoil-hat age, with trust in democratic institutions under pressure, even a large amount of added expense could turn out to be money very well spent. •