Nick Xenophon’s vote – does it hit Labor or Liberal harder?


Last week, I posted data showing the Nick Xenophon Team vote at the 2016 Senate election broken down by state electorate.

In the process of completing my guide to the South Australian state election, I noticed the trend in terms of which seats popped up with very high NXT votes. It appeared that numerous safe Liberal seats ranked highly in terms of NXT vote, while the safer Labor seats ranked more lowly.

It’s undoubtedly true that Xenophon takes votes from all parties – when his vote has been particularly high, both major parties have suffered big hits. But the vote does tend to be higher in electorates normally considered ‘safe Liberal’.

We saw this clearly at the 2016 federal election – the NXT vote was highest in Mayo, Barker and Grey, all places where the Liberal Party would normally walk all over Labor in a head-to-head fight.

And when I compared the 2016 NXT Senate data to the Liberal two-party-preferred vote in each new state seat, there is a trend where the NXT vote is higher in stronger Liberal seats.

It’s not a perfect trend, but the correlation is around 0.558.

So what does this suggest about the impact that Nick Xenophon’s SA Best could have on the next election?

We obviously don’t know how highly the party will poll – they have only announced a handful of candidates, so it’s conceivable they won’t run a full ticket. But it seems more likely that a large vote for SA Best will hit hardest in Liberal seats, particularly in seats where the Liberal Party would normally not need to campaign hard. In contrast, the safer Labor seats are likely to stay safe, unless SA Best polls particularly strongly.

Liked it? Take a second to support the Tally Room on Patreon!


  1. What are the chances that Labor run dead in safe Liberal seats? Preference SA BEST and then aim for third place so that the preferences count? I think they’d be mad not to.

  2. Makes sense. The hard right of the Libs are being very vocal and dominating the party, at least in the media, whilst the bulk of Lib voters would be much closer to the centre and would be desperately looking for any alternative other than the ALP.

  3. Xenophon makes a Liberal Government less likely, by dividing the right of centre vote and potentially becoming the biggest non-ALP party. This however makes it harder to have a state ALP government in the non-short term because it creates the potential for a non-ALP state government not tied to Commonwealth Coalition Governments, reducing drag on their vote.

    Xenophon would also have far less incentive to stick to the convention of deferring to the PM on half-Senate election dates, as SA best would not be in government at the Commonwealth level (except very occasionally propping up a minority government) and thus not be at risk from a by-election protest vote effect by not having half-Senate elections at the same time as House of Representatives elections. This would reduce the vote of whichever of the ALP or Coalition was more on the nose at Commonwealth level (usually the party/coalition in Government) and also reduce the vote of both compared to minor/micro parties and the Greens.

  4. Xenophon’s move means it is unlikely his party will have any further success in the federal arena.
    His move back into state politics could mean either two things, voters could be swayed to vote for him in larger amounts in the knowledge that he could govern the state, that being said it could work against him as he has built his reputation on being a protest vote within federal parliament.
    I’m more inclined to believe that his vote will increase from the last federal election due to the fact that Labor is passed its used by date and the Libs don’t look particularly appealing either.

  5. Federally Xenophon didn’t threaten any Labor seats, but were at worst 4% away from winning all of the Liberal seats; Grey, Sturt and Boothby could have been won by better preference flows, and Barker is a marginal seat. A stronger campaign and they could have replaced

    Labor’s best hope is to run dead in every safe Liberal seat, and focus entirely on their own marginals. Greens would also be wise to stitch up a deal; a good NXT->Green preference flow is their only chance at a lower house seat (if they can snowball to the top in seats like Adelaide or Unley).

  6. If you look at each state seat’s normalised Senate 3PPs (and its nominal 2PP holder, then Xenophon candidates would come top-two in 23 seats. Of those 23, in 5 seats a major party would have a “primary” of over 50% (so the seat would remain out of play barring weird preference decisions from the minors).

    Of the remaining 18, 12 are nominally Liberal and 6 are nominally ALP.

  7. Would it be possible to do a similar analysis to convert Queensland state results to federal results? Interested in LNP vs ALP vs GRN in Brisbane, Ryan, and Griffith (and other Brisbane seats but not as crucial), LNP vs ALP vs PHON in most other seats and LNP vs ALP in every seat (except Kennedy).

  8. Historically I don’t believe converting results between levels of government has had much predictive power. Turns out people actually do vote a bit differently depending on what issues are in play.

    The ECQ only publishes results to the level of first preferences by polling place and preference distribution by district.

    I asked if they’ll do the polling place to SA1 correlation, but the reply I got claimed that they do that already (they don’t, at least not publicly).

    For the inner-Brisbane 3PPs you might as well just collate the state polling places lying within the federal divisions and apply the district-wide micro-to-3PP flows.

    For the parts of the state where PHON came top 3 consistently (i.e. they ran and came top 3 in all state districts covered by the federal division) the same approach applies.

    Keep in mind that there are substantial ‘candidacy effects’. PHON only ran in about 2/3 of seats, for example; I think there are six state seats where the Senate figures had them top-3 but they didn’t stand a candidate.

    For ALP vs LNP 2PP, I think you’d be more accurate if you just grabbed the post-2016-election swing (statewide, if available, otherwise nationwide) from a reputable national poll or aggregate, and also account for the boundary changes once those are finalised.

Comments are closed.