Possums Pollytics

Politics, elections and piffle plinking

Howards Movements

Posted by Possum Comitatus on August 23, 2007


When analysing what makes voters tick, we often get carried away with satisfaction ratings, preferred PM levels, trends, counter trends and notions of who can manage what best. Yet, after all the navel gazing and the statistical analysis of the goat entrails, some things stand out a hell of a lot clearer than others.

If we look back over the history of the Howard government, we know that they enjoyed, like all other governments before them, a honeymoon period.

We also know that One Nation kicked them around the paddock, we know that the GST pissed a lot of people off, we know that Workchoices is about as popular as canned cheese and we know that Rudd seems to have had a fairly large effect.

So taking these obvious things, let’s see just how they affected the governments primary vote and the size of that effect.

To start with, I simply turned these issues into dummy variables which had a value of one for when that issue played out on the political stage, and a value of zero for all other times.

I defined the Honeymoon dummy variable (DUMMYHMOON) has being active for the first 12 months of the government. The One Nation dummy variable (DUMMYON) was active from April 1997 when the party was formed through to the 1998 Election. I defined the GST effect (DUMMYGSTEFFECT) as being active over the first 12months of the GSTs life.Workchoices was defined as starting in November 2005 when the legislation entered the Parliament and extending through to today. And finally the Rudd dummy started in December 2006 and continues through to today.

Nice and simple.

With these issues all neatly operationalised, I then regressed the government primary vote on a constant C and these dummy variables using the period of the Howard government.

The results of the regression came out as:

Dependent Variable: GOVPRIMARY

Method: Least Squares

Date: 08/23/07 Time: 10:03

Sample: 1996M04 2007M08

Included observations: 137



S. Error



































Adj R-sq


The top image is the graphical representation of these results with the red line being the actual government primary vote and the black line being the fitted regression estimation. The blue line shows the residuals from the equation. Also note that all of the variables used are highly statistically significant.

What this suggests is that without any of these things happening, the government primary vote would be at around 44 (which is pretty close to its long run average).

It suggests that the governments honeymoon period gave them an average 5 point boost over the honeymoon period. It suggests that One Nation pinched an average of just over 3 points from the governments primary vote. It also tells us that the GST reduced the primary vote by about 3 points as well.

The two important issues lately have been Workchoices and the Rudd leadership. The way these two issues are interacting with the governments primary vote numbers is a little intertwined and the magnitudes on each issue may be slightly out as a result – but not by any great amount collectively.

It suggests that Workchoices has taken over 2 points off the governments primary vote and Rudd has taken 3.5 point off it. Although Workchoices by itself without the Rudd dummy suggests that it is probably closer to 4 points, with Rudd being closer to 2 points…. But I’m keeping this simple so we won’t go there.

All up, these 5 issues explain over 60% of the movement in the government primary vote since 1996.

This analysis doesn’t take into account the inertial behaviour of the primary vote series so the size of the effects above are slightly out. But after adjusting for that behaviour (which makes everything a lot more complicated to explain), the above results are only out by a maximum of about half a point thereabouts.



AddThis Social Bookmark Button


19 Responses to “Howards Movements”

  1. Stunkrat said

    There appears to be a spike and sustained period above the average; it kicks in Q3 2004 and dies in the bum Q1 2005. I assume that it’s associated wtih Latham and the 2K4 election, and then it dies in the bum when Latham went spak.

    Also, you misspelt “navel”. You’d have us staring at passing ships.

  2. Possum Comitatus said

    That spike was Latham – I was thinking of throwing it in (actually did to check it out) but it was a bit of a flash in the pan and just clogged the graph up 😉

    Ta for the spelling mistake.

  3. jasmine_Anadyr said

    congrats on your (I’m guessing) unpaid gig with Morgan Polling – they seem very very fond of you.

  4. lurker said

    So you’re saying there’s an interaction between WC and Rudd but the combined total eefect of the two is still about 6% drop whether you consider them independently or combined?

  5. Possum Comitatus said

    Jas – its actually a comment by Alan H they’re referring to back in the comments section of one of the posts.Their link just sort of turned up one day in their releases, as did Gary Morgan here with a statement in the comments section and its been that way ever since.

    Lurker, that’s about right.The effect of Workchoices is overlapping with the effect of Rudd so it gets difficult (although not impossible) to separate them out exactly.Collectively they have an effect of about 6%.The only question is over the proportions of the effect for each.My feeling is that Workchoices is a slightly bigger effect than Rudd because of the dynamics of the way they play out when I measure them in different ways.

  6. sinden said

    Possum, your analyses are fantastic. I do look forward to them.

    Just a short comment about “Howard Movements” (apart from the anal interpretation).
    Don’t want to bog all this good stuff down with minutiae, but I suspect that the NSW Election had a bearing on the Federal poll results from Feb till April.
    In the leadup to the NSW Election, voters in that state were reminded (in no uncertain terms) of the risks with Workchoices. The problems with Workchoices had been well known long before the election, but Labor advertising during the campaign very much focused peoples’ attention. Workchoices was THE issue.
    Given that Federal polling must contain a very high proportion of NSW voters, this could be the reason for the very low Federal Govt numbers around this period. Workchoices has since slipped off the top shelf (Haneef, amongst others) and the Fed Govt has benefitted from this issue not being at the forefront of peoples’ minds. No doubt Workchoices will be back – as this is (and rightly so) the real bogey for the Coalition.

    Keep up the good work.

  7. lurker said

    The govt ads are doing a good job at keeping WorkChoices in people’s minds.

  8. Possum Comitatus said

    I agree Sinden, the NSW election was one gigantic anti-workchoiced advert.While in NSW it was obvious, in the other states the news nearly always described the NSW election process as an election where workchoices was a key issue (regardless of whether it was or not).

    I dont actually think Workchoices has slipped off the top shelf of concerns of the electorate, rather just off the top shelves of the news cycle – but I do agree when you say it will be back.And back in spades come the ALPs election advertising.

  9. The Doctor said

    it looks like you have agreed with eyeball assessment in “For Whom the Inertia Tolls” !

  10. Possum Comitatus said

    Doc, three years worth of data giving a linear trend with a t-stat of -12.9… statistically significant to better than the 0.0001% level.


    I think that gets it beyond the eyeball effects of a three month trend that has been killed stone cold dead in the last week 😉

  11. fred said

    A very humble and tentative suggestion.
    If I accurately understand what you have concluded in a previous post then I suggest you could add another colour shaded section to your graph.
    After the budget of 2005.
    When the Coalition vote begins it’s steep long term decline.
    perhaps THE underlying factor to all that followed thus far?

  12. Possum Comitatus said

    No need to be humble and tentative around here Fred – no one else is 😉

    That’s a good point.From the graph, we can clearly see the prolonged movement away from the government from the 2005 Budget period.Unfortunately we can’t use a dummy variable to measure it (when we do it turns out to be statistically insignificant).

    This type of dummy variable analysis essentially measures how political events changed the level of the vote over the period of the political event in terms of how far away that level was from the estimated vote given by the regression constant “C” (which represent what the vote would have been had the political events we measured not existed).

    So if we run a post-2005 budget dummy variable in the regression, over the period it would be measured (lets say for 6 months after the budget) the vote slowly moves from being above the level given by the constant C (43.85%) to being below it.So the average level of the vote during that time would be close to the value of C as the times it was above the level cancel out the times it was below the level.

    If we give our post 2005 budget dummy a larger timeframe to work over, say from the 2005 budget through to the present day, the workchoices and Rudd dummy variables take over in terms of explanatory weight when they come into the equation, again leaving the budget dummy as having only a very small effect – but this time it would be a statistically significant one.

    The post 2005 budget effect, because of its incremental nature and the way other events started to impact on the vote shortly thereafter, is unfortunately not a good candidate to be measured using this type of analysis.

    But you are right, maybe the 2005 budget effect was a really powerful underlying factor that allowed all of what followed to have the powerful impact it did, and maybe those following events actually had less of an impact than the data suggests because the post 2005 budget period was the real driver.

    I’ll have to give this some thought and see if I can come up with anything that can pull these relationships out.Many thanks fred.

  13. Leopold said

    In relation to the ‘2005 budget’ effect.

    In Newspoll, there was indeed a gradual ebbing of the Government primary vote through mid-2005.

    In ACN however, there is what might be more accurately called a ‘Beazley effect’, which happens in February/March after Latham is removed from the leadership. The Coalition’s primary vote drops from 52 in January to 47 in February and 41 in March. It stabilises at around 42-43, and then essentially stays put until September 2006 (barring a WorkChoices blip in November/December 2005). Only around half of that 10-point decline in Coalition support came back to Labor.

    Maybe Newspoll, because they don’t actually list all the alternatives but just ask ‘who would you vote for’, overstated the government’s position in the early days of the term – because people are not really switched on or inclined to be thinking about alternatives by themselves at that stage, and because people like a ‘winner’. The result of which would be the gap that opened up between ACN and Newspoll in the period between March and July 2005 in terms of Coalition primary vote.

    This of course reminds us of the idea that Latham cost Labor a winnable election in 2004, but I digress…

  14. Possum Comitatus said

    Hi Leo, its funny you should say that, I’ve been thinking that very thing this morning, both with the 2005 post budget effect and Latham.Great minds think alike!

    Or fools seldom differ, but we wont go there 😉

    I’m using Morgan at the moment to iron out some interpretations of the mechanics of the post 2005 budget effect because of the big Latham distortion that took 12 months to wash through the system.

    Latham really did cost the ALP the election.I’m looking at the quarterly newspoll breakdown at the moment, and there was a clear declining trend against the coalitions primary vote in their safe seats and their marginals until the april/june quarter 2004.It’s my initial reckoning that the government, in their primary vote, got a 4 point boost in the marginals and a massive 8 point boost (at least) in their safe Coalition seats because of Latham.The government have been in serious trouble in their safe seats for years, but Latham hid some of the problems.

    Those artificial boosts have now dissipated and returned the primary voting patterns back to their long run trends.

    I’ll fix this up today and tomorrow, gather my thoughts and whack it up on Monday – I’d be interested in your comments when I do.

    But from looking at this initially – this election is over.A minimum of 30 seats are going to change hands.

  15. KC said


    I don’t know whether Latham by himself cost the ALP the election.

    Can you have a look at the minor party votes, eg One Nation, Democrats and theorise where they went. These lost souls had to go somewhere.

  16. Tomasso said

    Hi Poss,

    Separating Workchoices impact and Rudd impact is probably the wrong way to go. Workchoices (and most of the other dummy vars) are policy or disposition effects (moods?). Rudd dummy (+/- whole party) represents the quality of the tactical play (by the ALP) on top of policy/disposition (as well as the “new boy” effect).

    Part of the “Rudd driven” (downward) movement for the LP would be the *interaction* with Workchoices. Specifically, how well the tactics were played (by Rudd et al). The other effects are WC on its own (pre-Rudd and carry-over), and Rudd (et al) on its own.

    I doubt you can tease these out, but maybe it can be done…


  17. Possum Comitatus said

    KC, it probably wasnt *only* Latham, but where the leader goes, the party follows.With the minors, the One Nation vote came somewhere between the 60/40 to 70/30 proportion from the Coalition/ALP. It picked up a few points of the other minors vote as well.When One Nation left the field, nearly all of the vote seemed to go back from where it came.The Democrats vote in the lower house hasnt really been big enough to measure where those voters went after the Dems vote collapsed – which is a pity as I too would love to know how that panned out.

    Hi Tom,
    I was of similar thoughts about whether to include Rudd for pretty much the same reasons (policy consequences vs basically PR effects), but Rudd had such a big effect on the polls instantaneously that it left me wondering whether it was more a result of tactical management of the “isnt he a nice boy” type effect or if Rudd just simply opened the door up on a whole lot of people that had been looking to vote for anybody but Howard for a long time, but couldn’t find anyone to give their vote to.

    If it’s the latter, then I dont think the nuances between Workchoices and Rudd will ever be able to be teased out, opening the door was as important to the size of the voter movement as the build up of pressure behind the door that workchoices would have contributed to in a significant way.

    The best I could do was just run them as separate variables, play around with each of them individually and give a bit of a blurb about how the magnitudes of the two effects might be a bit different, their combined effect would be pretty much the same.It would be nice to tease out the differences, but I think that’s a job for qualitative polling that, you never know, might turn up after the election.

  18. Possum,
    Enjoying this site and its analysis.
    Look forward to interesting days ahead.
    Myself, I think the proverbial drover’s dog has a good chance on polling day as Howard et al have offended so many over the last eleven years.
    For some it won’t be a vote for a particular candidate, rather a very personal vote against John Howard.
    Keep up the good work.

  19. Stephen T said

    Possum thanks again for great analysis. Your sight has been an absolute boon to those of us who are looking for hard headed analysis. It has been obvious to me that your work has clearly shown a substantial win to Labor for some time. I have a lovely little chuckle at all the pundits who are saying the polls are shortening. Oh my God! its nearly 50/50. Puke, blah blah. Cognitive dissonance writ large by a bunch of desperate try hards. Even the GG is struggling to draw anything meaningful from a dried up conservative corpse shriveling day by day due to Howard’s vampire-ism. Just love the solid trend.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: