Nicholson in the Oz

welcome to               Mumble   =

A Newspoll table: approval ratings, voting intentions and election results

Table below shows Newspoll data, over two decades, with average voting intentions and satisfaction with Prime Minister and Opposition leader over the various pairings. Last column shows actual election result. These numbers are only until that election, so don't include data after July '87, March '93 etc. (In all cases, an election is followed by a hike in successful PM's satisfaction and voting intention and dive in opposition leader's.)

Howard I, the '85 to '87 version, had both poor satisfaction ratings (31) and voting intentions. At the July '87 election he got 49.2, well above his average survey support of 47.1. Peacock from '89 to '90, a similar story. (Don't have data for Peacock's first stint.)

John Hewson, the opposite, a high approval rating and voting intentions lower than his survey data. Same with Mark Latham (highest opposition leader approval ratings in history), and the Beazley II, 1998 to 2001, kind of belong in this category. (Beazley I was well approved of throughout the term 1996 to 1998 but started it a mile behind in voting intentions. He overtook about half way through as the PM's approval went south. I may separate this out at a later stage.)

If there is a pattern, it is that voting intentions are more sustainable if accompanied by low (rather than high) approval ratings

Howard II, '95-6, had both high satisfaction and solidly high voting intentions and a high election result, the ideal combination (and probably an exception to this vague relationship, but a 'neutral' one: his voting intentions didn't get a large boost on election day).

Have separated Beazley III into two segments: the first 8 months and the next 11: in the first he was getting what the commentators today want: a high approval rating but low voting intention (eg Peter Hartcher compares Bomber unfavourably with Larry Springborg, who got quite high approval but voting intentions - and actual election result - in the mid-low forties), over the last 11 months the other way around. The Howard I and Peacock precedents would suggest Beazley's trend vote will improve as election time approaches. This is what I reckon will happen, assuming he stops trying to be more Howard than Howard (has Mark Latham joined his advisor team?)

Once again I pose the question: what does 'satisfaction with opposition leader' mean? You've got the 7 percent Green voters who can't stand Beazley, but will mostly give him preferences. Half the respondents are giving their two cents worth on the political scene. Whatever it indicates, it is not a direct, linear predictor of electoral success. 

Touched on some of this a few months ago. Will return to this table.

• Two party preferred voting intentions are calculated from primary support data. Newspoll has had several strategies for calculating notional two party preferred, none of them quite satisfactory.

• With individual polls I would round my voting data calculations to whole numbers, because published primary voting intentions are rounded. However, I feel publishing to one decimal place is appropriate when averaging a number of polls; can't put my finger on why but am sure a statistician would agree.

Sept 18 With 'mates' like these ... I love a claustrophobic country. Earth to Kim: the way to win from opposition is to neutralise your weaknesses and play to your strengths.

If you take up the PM's songbook and find the eggshell-thin, corny tunes not really your bag, you'll sound unconvincing and flat and make him seem better than he really is. But if you do actually dig them, and so belt them out with gusto (as Marky Mark did), that's even worse because you're doing all the work to make them more popular while the royalties flow to the composer. Either way you'll give the ditties added appeal but folks will still  prefer the originals.

The Association of Music Critics and its president Paul Kelly reckon the Howard repertoire represents the end-point of the evolution of this country's music. They say it is more 'Aussie' than all before it and anything that follows must build on it. But they're wrong: those songs just happen to be in the CD player at the moment.

A clunkety clunk metaphor I promise not to use again. 

Or as Michelle Grattan writes: "he needs not so much to play the hairy-chested tough man (a part he ill fits) as to actually be tough when that is required."

Having said all that ...

Peter Hartcher in SMH reckons 'approval ratings' (opinions on how the opposition leader is travelling) are more important than voting intentions (which way people reckon they'll vote). 

In a few days I'll put up a table that suggests a counterintuitive: a given opinion poll voting intention level is more sustainable if accompanied by a low approval rating than by a high one. Fine examples of each extreme were Andrew Peacock and Mark Latham.

Sept 13 Ballot paper order

What do Kim Beazley, Ross Cameron and Paul Neville have in common? According to Amy King and Andrew Leigh, they would all be ex-MPs (Cameron, who already is, would have become one 6 years earlier) were it not for the luck of the draw.

Sept 12 Opinion polls

Newspoll in Oz, 53 to 47. Trying to pin-points reasons for fortnightly poll shifts is nearly always pointless; Dennis takes an exceedingly long bow to last week's super changes.

Yesterday's Nielsen, no link, 52 to 48.

Bob McMullan on opinion polls

Happened to hear (on radio) Bob McMullan in Parliament yesterday in a grievance debate, making the point often made here, that the current narrative of an unbeatable federal government led by a political genius versus a hopeless opposition (not his words) determines the reporting of published opinion polls. Even though most (eg, see above) actually show Labor ahead, most people wouldn't know it. Not a partisan thing, not deliberate bias, and he's not sure it would serve Labor's purposes if surveys were reported properly. 

Bob also complained about absence of qualifiers such as error margins, which are routinely included in other countries' reports. He concluded:

over the months ahead I intend in parliament and outside to raise a number of issues about the state of our political system, our governance and our democracy. This has been just the first instance.

 Hansard extract here, starting bottom left hand column.

Sept 10 Qld election: from the tallyroom

The morning after

Dun all my Centrebet dough. At campaign outset I thought it might be close-ish, but to now blame Dr Flegg and campaign blunders would be too easy and wrong. It's probably the case that it was never going to be close, and even if the Coalition had put in a blinder of a campaign they would have done a little better, but that's all.

Eight years or three terms?

Further to pondering on time in office versus terms, let us compare last night's result with NSW in 2003, when an 8 year old Labor government went for re-election. Carr & co were looking shaky, not greatly loved, much grumbling about  health and the Sydney perennial of trains, a few dreadful byelection results under the belt. Below I've extracted, from the very accessible Newspoll website, two party preferred data for two years prior to the 8 year election of both Carr and Beattie governments. (Calculated Carr's two party preferreds from Newspoll primaries.) Red is Labor, blue Coalition.

Rather similar shapes, with government recovery beginning before campaign. (In Queensland a Newspoll published in first week was mostly taken before election called.) In both graphs the final data are actual election results, in both cases rounding to 56 to 44. 

Incumbency

As you know, I'm a relative disbeliever in things like 'issues that swung the campaign', 'momentum', 'cutting through' etc etc. I'm not even sure about 'having to present a case for change'. These are always defined after the result, with hindsight.

I think Michael Warby described incumbency as a 'diminishing asset'[?]. At the state level today it seems to have a long half-life. I do repeat myself, but in all states and territories (except WA) over the last decade Labor governments have been re-elected by huge amounts, in most cases setting records. Every state commentator will explain that the Premier is brilliant and charismatic and the opposition incompetent, 'not ready to govern'. Plus this and that issue.

In the last few weeks alone we've read lots about how the same state Liberal machines produce wonderful politicians federally but throw up total duds at the state level - and vice versa for Labor. I don't believe it, it does not compute in my overly left-brain.

Apparently, state Liberals 'lack charisma'. Well, watching Springborg and Flegg's concession speeches last night, and imagining they were victory speeches instead, I could not discern inherently less of the stuff than could be found in, say, a Carr, a Iemma, a Rann or, of course, a Howard. 'Charisma' is nearly always context dependent.

But having said all that, I believe Peter Beattie is the only politician around today with that special something, a warm and open communicator, a good faker of sincerity. Have I mentioned before that he should go federal (this year)?

Federal implications

Beating incumbents is still difficult. Bad news for Mr Beazley.

Sept 9 Qld Election day II

Snippets from Brisbane Town Hall polling booth this morning. Final photo, bottom right, is the empty tallyroom in Southbank, all systems tested and ready to go.

Sept 9 Qld Election day

• Newspoll 55 to 45, yesterday's Galaxy in Courier Mail 56.5 to 43.5, Mr Morgan 53 to 47.

• Video of yesterday's leaders' debate at the Courier Mail.

• Centrebet's seat betting odds (which mostly reflects punter sentiment) has been all over the shop. Now $1 will only get you $6 for Labor winning more than 63 seats, down from $34 two days ago. In general the assessments have flattened out, so that most individual seat bets between 50 and 63 pay more than they did a few days ago; most people think Labor will win but actual majority hard to predict. Bets will probably be no longer available. PDFed at 7:50am here.

Sept 7 An increased majority in Qld?

Centrebet has made a few adjustments to those below, including upping the 'more than 63 seats' odds to 33 to 1. [update: two days later collapsed to 5 to 1.] (Labor won 63 at the last election.) I think they are good odds. Beattie won more than 63 seats in 2001. As well, it's an eight year old state government going for re-election, and the last one of those was Bob Carr's in 2003 which [from memory?] increased its (notional) majority. These things can happen.

Sept 6 Punting at glug glug

Just put in my 89 probabilities. Phew, it takes a while. 

I PDFed it before submitting,  here. [Just noticed I've only given ALP 45% chance of retaking Redcliffe. This was an error, and I'm punting on all three by-election losses returning to Labor. Update: Mr Farmer has allowed me to change it.]

Was my first time looking at any of the seats; as you know, I favour macro over micro, so buckets of guesswork and split second punts. I think it adds up to 53 or 54 seats for Labor (counted twice, got two results); Centrebet's shortest odds are on 58. [Thursday update: Centrebet shortest odds now 56.]

Speaking of Centrebet, if you're feeling mad enough to bet on a Queensland Coalition win (as I was a few weeks ago), you should take the 20 to one odds on Labor getting fewer than 50 seats out of 89 (which also allows for a slim Labor win), rather than the miserable 9 to one on offer for outright result. [Thursday update: Centrebet has adjusted the 'fewer than 50' odds to 8 to 1.]

Personally (and don't try this at home, as my betting record is poor) I've put a few bucks each on: ALP wins fewer than 50; more than 63; and on 52, 53 and 54. That is, with the combination of three combustibles Queensland voters, OPV and Peter Beattie, it could fly off anywhere. Plus I saw Larry Springborg on telly last night, and he didn't seem too bad at all.

Sept 5 Speaking of preferences ...

(as we were last week), Parliament House Library has published 2004 federal election analysis. Not as ambitious as that for 2001.

A few points: in eight seats Labor trailed on primary votes but won after preferences (Coalition won none this way); Family First went 67% to the Coalition; and Greens flows to Labor have increased from 67 percent to 81 percent from 1996 to 2004.

Aug 30 Newspoll has a whopping lead for Qld Labor, 58 to 42

Have added Queensland links, above, which will stay until election day. Let me know of others. Courier-Mail's has a map, which is good, but it's not colour-coded, which is a pity.

Aug 29 Newspoll in the Oz

Says 51 to 49, from primary support of 42 to 44, 'based on the second-preference flows at the last election', writes Mr Shanahan. Presumably (and hopefully) that's not right, and they're still using full preference flows.

Aug 28 Family First preferences

I had a piece in Saturday's Australian Financial Review here

• [Update: Antony Green adds an important caveat to the several below: how-to-vote material displayed on voting screens at SA polling places. Read his email.]

Some extra bits/accessory items not squeezable into 500 words:

• A wee howler: I wrote that Family First didn't preference Labor anywhere in 2004, but I now understand they did in Brisbane, where the Liberal candidate was gay. (I thought they'd run a split ticket; it made no difference to the result.)

• Late last year I wrongly accused (in strong terms) the shadow foreign minister of telling fibs on the topic. This evidence suggests he knew what he was talking about. [Update: reader Dave reckons I'm now too kind to Mr Rudd, because even interpreting these numbers literally, it was not the presence of a Liberal-FF deal that cost Labor seats, but the absence of an ALP-FF one. Yes, I probably did over-compensate.]

• South Australia has an unusual House of Assembly ticket system which catches votes that would be informal in other CPV jurisdictions. (For explanation, go to fifth paragraph of Antony Green.) The numbers given are nett of tickets (a matter of a few percent in each case).

• One Nation voters were quite directable, but being seen to deal with the party was death for conservative governments in Qld and NT in 1998 and 2001; leafy urban Liberal (or LCP) voters reacted angrily. Playing footsie with Family First doesn't attract anything like the opprobrium.

• I'm sure Greens have preferenced the Libs at times, in Queensland for example, but I don't know the flow figures. My source for 3.5 number is this Federal Parliamentary Library analysis of the 2001 election (near bottom of page), which could only compare split tickets and preferences to ALP. A proper comparison would also include instances of Green prefencing Coalition.

• That aside, we might make a quick calculation: multiplying the FF 2.1% vote by 26% gives you .55%; Greens' 7.2 times 3.5% is .25%. So perhaps we can say (with a bucket of qualifications) that a Family First preference deal is worth twice a Green one for a major party.

• Table below has the most marginal seats won by the Coalition on 2004 in which Family First ran. The four seats in which 26% more of FF vote would have made the difference to seat outcome are italicised. 

• Seat	Labor 2pp 2004	FF vote	Labor 2pp + 26% of FF
  Kingston (SA)	49.9	5.6	51.4
  Bonner (Qld)	49.5	4.4	50.6
  Greenway (NSW)	49.4	1.2	49.7
  Wakefield (SA)	49.3	5.4	50.7
  Makin (SA)	49.1	4.9	50.4
  Braddon (Tas)	48.9	4.0	49.9

• Of course, preference decisions themselves affect support for the party doing the preferencing. See note about One Nation, above.