Possums Pollytics

Politics, elections and piffle plinking

A Jump in the Wayback Machine

Posted by Possum Comitatus on December 2, 2007

Way back in the annals of history, when the blog first started in May 2007, one of the first things I did was build a very rough election prediction model – and it’s probably worth going back to have a quick squiz at it because we’ll be dealing with it here. What was interesting about that model was that it was more explanatory rather than predictive and it looked at relationships between variables over the entire 11 year period of the Howard government.

As a result it turned up some pretty interesting results, the most interesting being that the standard variable mortgage rate over the first 6 years of the Howard government actually had a positive statistical relationship with the government’s primary vote, meaning that as the standard variable mortgage rate increased, so did the government’s primary vote. It wasn’t until the second half of the Howard government’s reign that the relationship reversed itself, whereby the relationship between the cash rate and the ALP vote became positive. Yet, over the entire period of the Howard government up to May, the relationship between the mortgage rate and the government primary vote was still positive, albeit getting weaker with every rate rise (because, as we know – for the previous couple of years the actual relationship between rates and votes had reversed itself)

Back in that very first election model, we forecast ahead using some assumptions about the variables involved. We assumed that the satisfaction ratings of the two leaders would remain the same (which they basically did) , and assumed that the interest rate to disposable income ratio would continue on its medium term growth path (which it basically did too).

I just went over the model again, but this time feeding in the changing values of the variables as they occurred. So as the satisfaction ratings changed each month, those new values were fed into the model, but I maintained the internally generated vote estimates as the lagged values.

As a result, if we compare the original May forecast, with the updating forecast and the actual Newspoll monthly average results over the period of May to the election we get firstly on the ALP primary vote:

decaplprimforecasts1.jpg

And secondly on the two party preferred forecast:

decapltppforecasts1.jpg

Currently, the AEC has the ALP two party preferred election result on 52.9% and the ALP primary on 43.7%

What I find amazing here is that this very early explanatory model projected forward produced far more accurate prediction results than the last model I did just before the election. Both the updating model and the original model in May that used the assumptions about satisfaction ratings, both ended up being close to the actual election result with the May model predicting 53.7 and the updating model predicting 52.8. The Newspoll monthly average for November ended up at 53.5

On the primaries, the May model ended up with 45.6, the updating model with 44.5 and Newspoll at 46.2. The updating model ended up being only 0.1% out on TPP terms, and 0.8% out on primary vote terms. The May model with the assumptions was only out by 0.8% on TPP terms and 1.6% on primary vote terms.

I’m actually quite amazed that it worked so well, considering it forecast out 6 months into the future.

6 Responses to “A Jump in the Wayback Machine”

  1. Bushfire Bill said

    I guess the lesson of this election and its polls is that statistical determinism is only part of the answer. It’s not all in the numbers after all.

    Elections are human enterprises and are thus subject to the foibles of the participants (presuming they are human).

    I think a lot of what Possum came up with regarding the drift away from the Coalition by the Howard Battlers (that drift being basically set in place with the elevation of Rudd to the leadership) was completely valid and exceptionally useful in understanding what was going on in the electorate, at least on a macro scale.

    But when it came down to it nasty advertising, big budget smears, gaffes blown out of all proportion and so on had their effect. There is no way that any of this could have been predicted as early as May, September, or even perhaps October. That the May curves intersected pretty well with the actual results I think is more serendipity than anything else.

    I first realised we were going to win around a late January campfire up the mid-north coast of NSW. All the parents, grandparents and thousands of kids from the bushland caravan park were there roasting marshmallows and getting eaten alive by mozzies. The verdict around that campfire was pretty-well universal: Howard was gone and deserved to go.

    I came away convinced that Labor would win the election if nothing went too wrong for Rudd. The mood was right. It was as good as any poll, at least at the time.

  2. Possum Comitatus said

    You’re right BB, it’s been an interesting exercise seeing what empirical ‘things’ influence the vote and trying to find out the size of that influence over differing time spans. It seems that the sweep of “big” issues like interest rates, housing affordability, leadership ratings and satisfaction ratings work over periods of years rather than months,and explain about 60% of the vote movement. Then those 13 million individual vibe issues make up the other 40% or so. It looks as if over the last few ears the big sweeping issues that could be measured put the TPP vote at a given level, then all the little things moved the true value around that given level by about 5% or so either way.

  3. Tomasso said

    I liked the way you managed lags in the early model. Lags let the mix of the population say a bit more, and give time for effects to settle in. I’m speaking in hindsight now, but I did like it the first time… Tom.

  4. steve_e said

    The Polls provide an insufht to those who are polled. There is a large number of people who will never be polled – not at home, no landline phone, etc.

    There is a percentage of voters who remain undecided normally through lack of interest or engagement in the election process who, because of a need to vote else be fined, use all sorts of processes in the voting booth. Any analysis of a narrowing is an attempt to guess what makes these people tick. Since we have no data, it is all guess work. From discussions I have had post election with people, a view that is mention often is: “If it ain’t broke, don’t fix it”. I am not sure what they mean by broke but from the context of the discussions, I assume this to be their income, their tax, the Family Tax Benefits, etc – in summary Middle Class welfare plays a big part of what makes these people tick.

    Hugh McKay’s interviews with people led him to conclude pre election that Howard would loose and loose Bennelong. He may have more insight into why this has happended because he has a wider penetration of the public than does any Election Poll.

  5. Ron Brown said

    How can we explain that when Beasley was Leader in October 2006 ,
    the Polls showed the Labor swing other than the ACTU work choices campaign
    ie. this pre -dated any 2007 cash rate rises

    (unless it is suggested that Morgan’s “soft” voter intentions have merit)

  6. Rod said

    Following on from Steve_e’s comment, I personally reckon that a significant part of any divergence between polls and election results occurs because of the behaviour of those who, for reasons of their own, are never prepared to take part in opinion polls. THis block of people remain completely un-assessable through normal polling processes. I doubt that they actually change their opinions much during campaigns, because I doubt they care enough to take any notice of them.

    In many cases they are probably people who are very disinterested politically, and on balance probably simply vote for the incumbent on the day, but ,barring the introduction of some innovative technique (paying for survey responses perhaps, though this would introduce other problems of its own) we will never really know what they are thinking.

    Changes in poll results DURING a campaign may actually also be due in part to increasing engagement, and therefore a greater preparedness to participate in polls, as the election date approaches. At other times the importance of cooking dinner or watching Neighbours outweighs any interest generated by responding to a survey. There is no reason to believe that any such factors will be resolved by processes of randomisation. The “block” of politically apathetic may well be skewed towards one side or the other.

    On a different note, looking at the State results, I see that the only parts of Oz where the 2PP Labor vote was less than 54% were WA and Queensland. I wonder how much differing poll performance is related in part to variations in emphasis in polling in different states. Seem to remember some of the ACN polls, for example, were heavily NSW / Vic focused.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

 
%d bloggers like this: