Late counting in the Melbourne byelection hasn’t changed the overall shape of the result (Labor has a narrow but adequate lead of 770 votes), but it has at least pushed the turnout figure above two-thirds, to a still-woeful figure of 67.43%. In an electorate with almost 45,000 enrolled voters, about 14,600 of them failed to show up.

Two main explanations seem to be offered for this. One is a general apathy and disillusionment among the electorate, particularly young people; the other is the absence of a Liberal candidate, supposedly leading to a lot of Liberal voters staying home.

I think there’s an element of truth in each of these, but even taken together they don’t explain the large drop in turnout.

Consider the Liberal voters first. At the 2010 general election the Liberals had 28% of the vote; on Saturday the three fundamentalist candidates and two broadly Liberal-aligned independents (Stephen Mayne and David Nolte) won 15.5% between them. Add in some more going to the donkey vote, the informals and the range of other independents, plus the 3.7% increase in the S-x Party vote, and you’re clearly not very far off the expected level of Liberal support.

Moreover, if it was mostly Liberal voters staying home, you’d expect the turnout to have dropped more in the strong Liberal booths and less elsewhere. But there’s no such pattern. In East Melbourne, for example, the strongest Liberal booth in the electorate (and where — disclosure — I was scrutineering for Mayne), the vote was down 2.9% on 2010, only a little more than the 2.3% average across the booths.

What about general apathy — are byelections generally showing a disengagement from politics? To some extent they are, but Melbourne is very much an outlier. Here are the figures for the past few years worth of byelections across Australia, showing the turnout compared to the previous general election and also whether or not there was a contest between the major parties:

A drop of somewhere about 10% is normal; it tends to be less when both major parties are standing, but not by a great deal. Certainly the list includes several that, on anyone’s account, were more boring than Saturday’s contest, but where the turnout still held up much better.

One, the South Brisbane byelection earlier this year, recorded a turnout almost as low as Melbourne (indeed with counting of late postals the Melbourne figure may still edge ahead). No doubt that was partly due to a fatigue effect, coming only a month after the state election with the retirement of former premier Anna Bligh.

But it’s surely significant that South Brisbane and Melbourne are similar sorts of electorates: both inner city with lots of young voters, particularly students, who are disproportionately likely to move around without necessarily bothering to update their enrolment details.

In a general election that’s less of a problem, because even if (for example) they’re now living back with their parents in the suburbs, they can turn up to a polling place there, discover that they’re still enrolled for the inner city, and fill out an absentee vote. (If they’ve permanently moved then there’s a fraudulent aspect to this sort of transaction, but no doubt it still happens.)

But in a byelection there are no polling places outside the electorate, so no absentee votes. In 2010 the absentees in Melbourne amounted to about 4500, or 10.2% of the enrolment. That must be a big part of the explanation for the drop in turnout. And since those young voters are more likely to vote Green, it also helps to explain Labor’s somewhat unexpected victory.

Further confirmation comes from the outlier at the other end of the above table, the Fremantle byelection of 2009. It defied the precedents by actually recording a higher turnout than the previous general election. But because it was held in conjunction with a statewide referendum on daylight saving, absentee voting was available. (Even taking that into account the byelection turnout was very good, but no longer so exceptional.)

That’s not to deny that voter disengagement is on the rise. Byelections in previous decades haven’t shown such steep decline: to pick one example at random, the Corio byelection of 1967 produced a turnout of 93.4%, down from 95.9% at the previous general election. But beware of putting too much faith in the simple explanations.