Seat polls – how do they compare?

7

We’ve seen a surge in seat polls at this election.

By my count I’ve found at least 50 polls which have been publicly released since the beginning of April. This is suggests we are on track for a big increase since 2013. You can download my list of these polls here. William Bowe lists 58 polls in the whole 2013 campaign.

This includes at least twelve seat polls from Newspoll released this morning, although the majority of polls have come from Reachtel.

Others know more about the problems inherent in individual seat polling than I do, but it appears to my eyes that part of the problem is that most seats only get polled once – so we don’t have a sense of a trend, or whether the poll is an outlier. This time around, at least ten seats have been polled at least twice, and in Lindsay we have four publicly-released polls.

The conventional wisdom is that, while Labor is neck-and-neck in the national polls, local swings will stop them, as the Coalition is doing better in their marginal seats as their first-term MPs benefit from new personal votes. This is undoubtedly true to some extent, but I thought it would be worth comparing local polls in each seat to what the state and national swings suggest.

I have taken William Bowe’s latest BludgerTrack estimate, which gives the Coalition 50.5% after preferences, as well as the state 2PP breakdowns, and plugged them into Antony Green’s House of Representatives calculator.

If you take the national swing, you get a result of 82 Coalition, 64 Labor and four others. If you take the state swings, you get a result of 78 Coalition, 68 Labor and four others.

In the following table, I have compared the most recent seat poll for the twenty-nine seats which have produced polls with Labor-Coalition 2PP figures. I have excluded seats where the Nick Xenophon Team, the Greens or an independent are in contention (so no SA seats are included).

Seat Margin Nat’l swing State swing Latest poll LNP +/-
Banks (NSW) LIB 2.8% LIB 0.5% LIB 1.2% 50/50 (Galaxy, 11 May) -1.2%
Bass (TAS) LIB 4.0% LIB 1.8% LIB 6.5% 52% to LNP (Newspoll, 13 June) -4.5%
Bonner (QLD) LNP 3.7% LNP 0.7% ALP 1.4% 56% to LNP (Reachtel, 9 June) 7.4%
Braddon (TAS) LIB 2.6% LIB 0.3% LIB 5.0% 53% to LNP (Reachtel, 14 May) -2.0%
Brisbane (QLD) LNP 4.3% ALP 0.2% ALP 2.3% 51% to LNP (Newspoll, 13 June) 3.3%
Burt (WA) LIB 6.1% LIB 3.1% ALP 1.7% 52% to ALP (Newspoll, 13 June) -0.3%
Capricornia (QLD) LNP 0.8% ALP 1.5% ALP 3.6% 50/50 (Newspoll, 13 June) 3.6%
Corangamite (VIC) LIB 3.9% LIB 1.7% LIB 1.8% 51% to LNP (Newspoll, 13 June) -0.8%
Cowan (WA) LIB 4.5% LIB 1.5% ALP 3.3% 50/50 (Reachtel, 9 June) 3.3%
Dawson (QLD) LNP 7.6% LNP 4.6% LNP 2.5% 50/50 (Reachtel, 7 June) -2.5%
Deakin (VIC) LIB 3.2% LIB 0.9% LIB 1.0% 52% to LNP (Reachtel, 9 June) 1.0%
Dobell (NSW) ALP 0.2% ALP 2.4% ALP 1.7% 51% to ALP (Reachtel, 9 June) 0.7%
Dunkley (VIC) LIB 5.6% LIB 1.1% LIB 1.2% 52% to LNP (Newspoll, 13 June) 0.8%
Eden-Monaro (NSW) LIB 2.9% LIB 0.7% LIB 1.4% 51% to LNP (Reachtel, 19 April) -0.4%
Franklin (TAS) ALP 5.1% ALP 8.1% ALP 3.4% 54% to ALP (Reachtel, 14 May) -0.6%
Gilmore (NSW) LIB 3.8% LIB 1.5% LIB 2.2% 51% to LNP (Galaxy, 11 May) -1.2%
Hasluck (WA) LIB 6.0% LIB 3.0% ALP 1.8% 53% to LNP (Reachtel, 16 June) 4.8%
Herbert (QLD) LNP 6.2% LNP 3.2% LNP 1.1% 54% to LNP (Newspoll, 13 June) 2.9%
Leichhardt (QLD) LNP 5.7% LNP 2.7% LNP 0.6% 52% to LNP (Galaxy, 13 May) 1.4%
Lindsay (NSW) LIB 3.0% LIB 0.7% LIB 1.4% 52% to LNP (Newspoll, 13 June) 0.6%
Longman (QLD) LNP 6.9% LNP 3.9% LNP 1.8% 50/50 (Reachtel, 2 June) -1.8%
Lyons (TAS) LIB 1.2% ALP 1.0% LIB 3.7% 51% to LNP (Reachtel, 14 May) -2.7%
Macarthur (NSW) LIB 3.3% ALP 1.1% ALP 0.4% 50/50 (Newspoll, 13 June) 0.4%
Macquarie (NSW) LIB 4.5% LIB 1.5% LIB 2.2% 50/50 (Reachtel, 19 April) -2.2%
Page (NSW) NAT 3.1% NAT 0.8% NAT 1.5% 56% to ALP (Reachtel, 19 April) -7.5%
Reid (NSW) LIB 3.4% LIB 1.1% LIB 1.8% 51% to LNP (Galaxy, 11 May) -0.8%
Robertson (NSW) LIB 3.1% LIB 0.8% LIB 1.5% 51% to LNP (Newspoll, 13 June) -0.5%
Wentworth (NSW) LIB 18.9% LIB 15.9% LIB 16.6% 58% to LNP (Reachtel, 31 May) -8.6%

Most of the seats on this list have polls in the 50-52% range – only seven seats had marginals above 52%. Things bounce around in that range, but the margin of error suggests that all of these seats could still be in play.

The interesting diversions include:

  • Bonner – the national and state polls suggest a very close race, but the LNP is winning comfortably.
  • Page – the national and state polls suggest a narrow win for the Nationals, but Labor is on 56%.
  • Tasmania – seat polls in the three Liberal seats in Tasmania all project a closer race than the statewide average, which suggests comfortable re-election for the three first-term MPs.
  • Hasluck – the most recent poll in the sample suggests a small swing against Liberal MP Ken Wyatt, despite state polling in WA suggesting a huge swing to Labor.

I also wanted to run through the trends in seats which have had at least three polls conducted:

  • Bass – Reachtel polled Bass twice in May – the first poll gave the Liberal Party 51%, the second favoured Labor with 51%. Newspoll polled there last week, and the Liberal Party had a lead of 52%.
  • Corangamite – this seat has been polled three times in the last three weeks. The first poll on May 26 had the Liberal Party on 54%, but this dropped to 51% in the two polls conducted in June.
  • Dobell has been polled once a month since April. Labor won 51% in April and 51% again in June, but in between in July the poll was 50/50.
  • Lindsay has been polled four times since April. Labor led with 51% in a Reachtel poll in April. The Liberal Party won easily with 54% in a Galaxy poll in May and a Reachtel poll on June 9, and this dropped to 52% in a Newspoll last week.
  • Macarthur – Labor won 51% after preferences in Galaxy and Reachtel polls in May. A Newspoll last week had the seat as 50/50.

I don’t think this answers the question about whether Labor or Coalition is on track to win. I still think the Coalition is more likely to win than Labor, and Labor needs a higher two-party-preferred vote, but the local seat polls in general confirm the national trend – Labor is just behind, but could win if enough seats break their way.

PS: I’ve started adding seat polls to the seat guides – they’re underneath the assessment section, and above the 2013 election results. So check them out next time you’re reading one of the seat guides.

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

7 COMMENTS

  1. Ben, I wonder if you might put up a post a day before election for everyone to out their guesses of final seat numbers. I’ve done guess for every seat before, but just guessing final numbers by party is just as good and easier.

  2. I also wonder if after the election you might do a post on electoral reform. How the changes went this time, how you think they will go next time, what would be the ideal changes to make, what are more likely changes to happen and what you think will actually happen?

  3. One thing to bear in mind when looking at these polls is that with so many polls, the chances of some of them being “rogue” goes up substantially – in the 50 polls seat polls we’ve seen, you’d expect 2 or 3 to lie outside their 95% confidence interval.

    It would be interesting to see the sample size alongside the result, if it’s known.

  4. @kme, sample sizes are in the spreadsheet I linked to at the top of the post.

    @Andrew, I usually do a predictions post just before the election. And I write about electoral reform a lot, will see what comes up.

  5. The newspoll sample sizes seem to be smaller….. the results tend to favour the lnp
    I understand the margin of error is 4.5%

Comments are closed.