Trump was not a significant factor, and that matters

Post election there is always jockeying to write the narrative. Frequently these are either complete fantasy, or “faction” – fiction based on fact – giving psychological support to the narrator, or manipulating listeners to the narrator’s advantage.

One theory that is gaining traction is that Donald Trump caused this result. It’s an easy story to write, given some added credibility by the election result in Canada and doesn’t require tediously poring over electoral statistics, talking to voters or analysing the evidence.

Two groups have a vested interest in hyping the theory that this was Donald Trump’s doing: the architects of the LNP campaign wishing to deflect attention from real faults and failings; and Greens, Labor and Independents wanting to delegitimize concerns about high immigration, government efficiency, free speech, DEI, education, government regulation, budget deficits, runaway debt, and lack of defence preparedness, that might be shared with Trump.

I’ve taken an early peek into the entrails of my qualitative exit polling and it shows Trump had very little to do with Peter Dutton’s loss. As of Sunday I had 578  responses.

The sample is left-leaning with Labor marginally over-represented, while Greens and Independents are significantly over-represented, and the Coalition is significantly under-represented.  

Out of the total sample 104 mention Trump in some way. This is 18%. But out of this subset Trump is overwhelmingly mentioned by ALP (42%), Greens (21%) and Independents (24%), a total of 87% of the sample.

Sounds like Trump might have been a hugely persuasive meme.

But not so fast. Are these habitual Labor, Greens and Independent voters, or have they changed their vote on the basis of Trump or Make Australia Great Again? They might even be fans of Trump captured by Clive Palmer’s Trumpet of Patriots tribute band.

A rookie error in analysing polling is to assume that quantity indicates that something was a deciding factor. To be a deciding factor it must move votes. If it is a clear concern for just one group of voters but not others it may rather be a tribal marker, but not significant to the whole community.

We ask people how they voted last election, and if they regard themselves as traditionally voting for one party or another. This allows us to identify genuinely swinging voters who are the voters who determine elections.

There was only one person who had changed their vote from the Coalition and cited Trump as a reason: “Mostly vote LNP. No longer.  Too many Trumpist crazies in it for my liking”.  

So one moving against the LNP out of a sample of 578 does not explain a swing of 2.56%.