Was it a protest vote?

The protest vote had nothing to do with this election. It's a bum rap. Some recent elections in Australia have been won on the basis of protest votes, but there are clear fingerprints when this happens. I've dusted the evidence and the fingerprints aren't there.

(Note: I've had to revise this article - see Whoops below) The fingerprint that denotes a protest vote is that there are larger swings in safer electorates than marginal ones. This has the result that seats which appear fairly safe are lost while marginal seats are often held; and the government can survive the swing with less than 50% of the vote.

By protest vote, I mean a vote which is cast to protest against the party that you expect to form the government, and that you want to be the government.

The graph below plots swings against margins. It views margins from the government's point of view. A negative margin is a positive margin for the opposition. The same applies to swings.

The trendline is a 3rd order polynomial. It's just a rough fit, but it shows that the data doesn't fit the fingerprint. If it did the graph would be a fairly gentle parabola on a north-south rather than east-west axis.

While 59 percent of voters expected Labor to win (according to Newspoll), voters voted on the basis of local factors with swings to the government occurring on a seat by seat basis. This suggests that while there was a definite swing against the government in a number of states, the size of that swing was dictated by a desire to change the government, not just send them a message, as well as local campaigns.

I've eyeballed the Greens vote and I don't think the relationship will be there either. But I'd need to look at it a lot more closely, and no-one gives me an easy way of downloading their data. The statistical relationship probably won't be easily seen, if it exists, because Greens only do well in a certain type of seat and service niche interests.

Protest_Vote_Fingerprint_Scatter_Graph_thumb_500px

 

It's not surprising that the protest vote isn't there. Both sides were saying it was close, and Tony Abbott was prefacing each issue with the injunction that "if you want to change "x" then you have to change the government". No chance of him slipping into office, as Kevin Rudd suggested he might, but what if Abbott had been more strategic?

Whoops

I've had to revise the graph above. Never do this stuff too late at night. When you plot the data on different axes you do indeed get a gentle parabola as a trend line. It's just too gentle and shows the tendency to vote against the government was a bit higher in non-Labor electorates than in Labor ones. So there is a protest vote tendency there, but a very small factor.

protest_vote_fingerprint_scatter_graph_correct_thumb_500px

Share this article on your favourite social bookmarking sites:
Digg! Reddit! Del.icio.us! Google! Facebook! StumbleUpon! Twitter!