Antony Green's Election Blog

Hung Parliament - Where to From Here?

Last weekend Australians thought they were voting on who would form government. If Labor or the Coalition had won a clear majority, this would have been a reasonable summary of what the election was about.

But with neither side having gained a majority in their own right, the murky world of government formation under our system of unwritten constitutional conventions has been exposed to the light.

Is Tony Crook, new Nationals MP for O'Connor, a member of the Coalition?

Since election night I have received more than 50 e-mails from members of the public wanting to know why I have included Tony Crook, the new Nationals MP for O'Connor, in the total of seats for the Coalition.

Mr Crook, like every other WA National candidate, was nominated under the umbrella of the Federally registered National Party. He appeared on the ballot paper with the party affiliation of 'The Nationals', as did all National Party candidates in New South Wales, Victoria and Western Australia. As far as party registration with the Electoral Commission is concerned, Mr Crook is in the same party as National MPs from other states.

If after the election Mr Crook or the WA Nationals no longer wish to be treated in this way, I can say on behalf of the ABC we are prepared to consider instructions from Mr Crook that he does not wish to be included in the total of seats for the Coalition.

If we receive such instruction to remove Mr Crook from the total of Coalition seats, we will take such action and ensure that it receives appropriate news coverage.

Distributing Preferences in Denison

In the last two days I have been deluged with questions about Denison. The main one is how can Andrew Wilkie win the seat if the Liberal candidate is finishing second?

The answer is that preferential voting is not a run-off between the two leading candidates. The count is conducted by succesively excluding the lowest polling candidates in order of ascending votes at each stage of the count.

Galaxy Marginal Seat poll

News Limited Sunday papers are this morning reporting a 'monster' poll of 4,000 voters by Galaxy. Galaxy has polled 200 voters in four key marginals in each state. The following table summarises the result.

Galaxy Marginal Seat Poll - News Limited Sunday Papers 15 August 2010

State (Seats)	2007 2PP	2010 2PP	Swing
NSW (Eden-Monaro, Gilmore, Macarthur, Macquarie)	50.9	48.5	2.4 to Coalition
VIC (Corangamikte, Deakin, La Trobe, McEwen)	50.4	52.0	1.6 to Labor
QLD (Bowman, Dawson, Dickson, Flynn)	51.4	46.0	5.4 to Coalition
SA (Boothby, Grey, Kingston, Sturt)	49.0	49.0	no swing
WA (Cowan, Hasluck, Stirling, Swan)	49.6	47.5	2.1 to Coalition

On Saturday evening (14 August), this was being reported as a poll showing the Coalition ahead 51.4% to Labor 48.6%. See the screen grab below from Skynews.

(UPDATE: See comments at end of this post. The Sunday Telegraph itself hasn't given any prominence to these national figures.)

When I entered the swings into the ABC calculator, I got the following result 

http://www.abc.net.au/elections/federal/2010/calculator/?swing=state&national=0.3&nsw=-2.4&vic=1.6&qld=-5.4&wa=-2.1&sa=0&tas=0&act=0&nt=0&retiringfactor=1

This produced a national 2PP figure of 51.1% for Labor, completely at odds with the figure of Coalition 51.4% reported above. Then it dawned on me that whoever calculated 51.4% has made a very basic error of political analysis.

The five 2PP figures for 2007 and 2010 reported above apply only to the four electorates in each state. They are not 2-party preferred figures for each state, and the Galaxy tables have been very precise in setting out the 4-seats per state nature of the poll.

What someone has done is take the five entries in the 2010 2PP column and average them to get a national figure. Wrong. Very wrong.

There are two serious errors commited here. First, the figures are for four electorates, NOT the states. Second, while the state samples are the same size, the state populations are not. To get a National 2PP figure, you need a weighted average of the state swings, NOT a simple average of the survey 2PPs by state.

The method I used in the calculator is the correct way to produce a national 2PP %. If you assume the swings in the four states are uniform in each state, the new state Labor 2PP %'s are NSW 51.3%, VIC 55.9%, QLD 45.0%, SA 52.4% and WA 44.6%.

These produce a National Labor 2PP % from this Galaxy poll of 51%, not 48.6%. Someone has made an absolute howler in trying to turn polls in 20 marginal seats into a national figure.

UPDATE 1: I was the victim earlier this morning of a headline that converted my post into a headline suggesting "The Galaxy poll is wrong, very wrong". I've had that corrected as it is not what I am suggesting. What was wrong was the headline national figure reported last night, and I see nothing to suggest Galaxy produced that figure.

As has been shown by state breakdowns of Nielsen and Newspoll, Galaxy suggests that Labor can lose the election in marginal seats in Queensland and NSW. The Galaxy poll is very much in line with what Nielsen and Newspoll have been reporting for three weeks.

However, the way the incorrect national figure was reported last night made this Galaxy poll look like some shock result. If it had been reported last night as a 1.7% swing against Labor costing it the key marginal seats it would have been a repeat message of the other polls. But instead it was reported with an incorrect Coalition 2-party result of 51.4%, which made it look like a shock departure from other polls.

UPDATE 2: Now I know about the print versions of the Sunday Telegraph, I know that the paper itself hasn't highlighted the national figure. It has treated the analysis properly as a seat by seat comparison. What appears to have happened is that the press release to several media outlets included a list of the seats and state swings and a total figure for these seats that was listed with caveats. That number has been taken out of context and turned into a television graphic attributing it as a national figure. As I thought, stuff-up rather than conspiracy.

I'm now going to close comments on this post as all that is coming now is endlessly repeated fanciful News Limited conspiracy theories and personal abuse directed at me. I have better things to do on a Sunday than try and moderate abuse.

Estimated Margins for Victoria's Proposed Federal Electoral Boundaries

If you are only interest in the 2010 election, pay no attention to this post. It is about estimated margins for new electoral boundaries that won't apply until the 2013 election.

To general surprise, proposed electoral boundaries for Victoria were relased on 30 July, right in the middle of the election campaign. The major proposed changes were the abolition of the safe Liberal seat of Murray in the state's rural north, and the creation of a new safe Labor seat reviving the name Burke covering areas along the Calder Highway between Melbourne and Bendigo.

Knock on changes have seen both of Labor's most marginal seats, Corangamite and Deakin, become marginal Liberal seats. The marginal Liberal seat of McEwen also became a safe Liberal seat, while the eastern Melbourne seat of Aston became much more marginal for Labor.

Interactive Map Application Launched

We've added another feature to the election site in an interactive map. You can find it at http://www.abc.net.au/elections/federal/2010/map/

It works like most Google interactive maps with the zoom strip in the top left, and the ability to grab and move the map by holding down the mouse pointer. Whenever you hover over an electorate, an information box appears telling you the name of the electorate, which party holds it, the name of the sitting MP and the margin. The box pops up next to the pointer at the moment which can be distracting. We'll try and tie the info box down to one spot on the page in the next day or two.

Another feature is that if you click an electorate on the map, the Green Guide information page for the electorate will appear under the map.

You can find the map by hovering over the 'Interactive' pull down menu on all the election pages. To get rid of the map, hover over the 'Green Guide' pull down menu and select another page from the site.

Have fun!

A Comment on the High Court Close of Rolls Decision

First a mea culpa. On ABC News 24 I think I just said something like the 1983 and 2007 elections were the only ones in six decades where the rolls closed when the election was announced.

Correct about 1983 but not true about 2007. Yes, the rules the High Court has now invalidated were in place. But the Howard government announced the election on Sunday 14 October. As a 6-week campiagn had been announced, writs were not issued the next day, but on Wednesday 17 October and on that day new enrolments closed, and for reasons to do with the Flinders Island Show Day, the roll for changes of enrolment did not close until 23 October.

The point I was trying to get across about the close of rolls is that it has been traditional for a period of time to be allowed between the calling of the election and close of the rolls. This used to be done by announcing the election ahead of the issue of the writs, with the rolls closing on issue of the writs after a period in which voters were able to regularise their enrolments.

This was changed in 1984 so 7 days were allowed after the issue of writs, the Howard government removing the 7 day period ahead of the 2007 election.

The High Court has invalidated the Howard government's amendment, which in effectively re-instates the 7-day rule. The AEC is now trying to deal with the enrolments received between the old close of roll date on 19 July and the re-instated close of roll date on 26 July. The AEC has estimated there may be as many as 100,000 but they won't know how many until forms have been processed.

(I explained the situation better in a previous post which you can read by following this link.)

Senate Calculators

Calculators based on the group ticket voting preferences are now live at http://www2b.abc.net.au/Elections/View/SenateCalculator.aspx?e=10&ca=nsw.

You can click through to all other states from the NSW page. Enjoy!

All the group voting tickets have also been published at http://abc.net.au/elections/federal/2010/guide/gtv.htm

For those of you who want to vote below the line on the Senate ballot paper so that you can decide who gets your preferences, two handy tools have been published.

The first allows you to choose groups on the ballot and then re-arrange them and print an example of a ballot paper that you can take along as a voting guide. You can find it at

http://belowtheline.cc/

Another useful tool takes the parties' registered tickets and allows you to re-arrange them, and again, gives you a sample ballot to take along.

https://www.belowtheline.org.au/

Let me remind you of two things.

First, you can't put the sample ballot papers produced by these sites in the ballot box. You have to transcribe the numbers on to the Senate ballot paper you are given. Only official ballot papers signed by a polling clerk can be used to vote, so you are still going to have to write in the numbers.

Second, remember to fill in all the numbers. There are odd savings provisions that allow Senate ballot papers with more than 90% of numbers and a limited number of errors to stay in the count as formal votes. However, to try and engineer your vote to exhaust its preferences requires a really good legal opinion over the meaning of Section 270 of the Commonwealth Electoral Act. Don't try and be cute, number all the squares.

Demography and Voting

Over the next week, I'm going to post some analysis of voting patterns based on comparing electorate based election result with selected demographic statistics taken from the 2006 census.

My source for the election data is the AEC's estimated votes and 2-party preferred percentages by electorate. I have taken the Census data from the "Electorate Division Rankings: 2006 Census (2009 boundaries)",  Research paper 2009-10 no.18, by Paul Nelson of the Statistics and Mapping Section of the Commonwealth Parliamentary Library.

My first analysis, on the relationship between 2-party preferred vote and the proportion of an electorate's population born in non-English speaking countries, ahs been published at http://www.abc.net.au/elections/federal/2010/guide/demographics.htm. More tables are to follow. The graphs on the page use Adobe Flash to interact with your mouse, displaying electorate names and values for points. You can also click on any point and link through to my Green Guide page for each electorate. Apologies to iPad and iPhone users, but you only see a static image.

Enrolment by Electorate for 2010 Election

A total of 14,030,528 voters have enrolled to vote for the 2010 election, an average per electorate of 93,537 voters.

You can find all figures via this link to the AEC webiste.

However, there is some variation in enrolment from state to state, as shown in the following table.

Summary of Nominations.

The number of candidates contesting the House of Representatives has fallen from 1054 in 2007 to 849 in 2010. The number of Senate candidates is down from 367 to 349.

The following tables summarize all the nominations by party, electorates by number of candidates and summary of Senate candidates and groupings.