Does Economic Management Influence The Primary Vote?
Posted by Possum Comitatus on October 17, 2007
A new day, a new Newspoll and a new take on the old hyperbole.
Todays Exhibit A comes from good old Dennis over at the The Oz:
“The Coalition has stretched its commanding lead over Labor on the key vote-changing issues of the economy and national security.”The problem with these types of claims is that they can actually be statistically tested.
Either fortunately or unfortunately for Dennis (I have a suspicion which), we can’t test the national security issue for there aren’t enough observations to get a decent statistical grip. But we can test for how economic management changes the vote.
Newspoll runs two polling series on this economic management question, the first is “which party do you think would be best to handle the economy?”, but there are only 11 observations for that series, which also happens to be the question asked today.
The other polling series asks “Which leader do you think is more capable of handling Australia’s economy?” The difference between the results in these two series is nothing for the ALP, with both series returning the same results for similar dates, within the margin of error of each poll.
For the Coalition, the former question produces slightly lower results than the latter question. However, the first series also has a far larger amount of undecideds. So the second series, the “which leader is best able to handle…” series should have more sensitive voters in it, and that is reflected in the larger variation in the second series than the first. This is actually really good news. Variation increases testability.
If we head over here to Newspoll, and press “Display results” on the question of “Which leader do you think is more capable of handling Australia’s economy”, a lovely little table pops up containing 47 usable observations of data that can be tested against the relevant monthly Newspoll aggregate of primary votes for those dates.
First we’ll run a simple regression using the primary vote of the ALP as the dependent variable, and the “Which leader is more capable of handling the economy” results for the ALP as the independent variable .What we’ll essentially be measuring is how much does the primary vote of the ALP change for every one point increase in the number of people that believe that the ALP leader is better at managing the economy. What we expect to see is a positive relationship that is also statistically significant.
So running the regression we get:
What this tells us is that yes indeed, the ALP leader having a higher result in the “more capable” question does in fact walk hand in hand with a higher Labor primary vote. With a coefficient of 0.196, we can say that this result suggests that a 5 point increase in the “more capable” question leads to a 1 point increase in the primary vote. We can also say that this is highly statistically significant at less than the 1% level (with a p-value = 0.0014) and that the “More capable” results explain roughly 20% of the variation in the ALP primary vote figures.
Dennis, mate…..it’s alright you can breathe easy.
For the moment 😉
If we do exactly the same thing with the Coalition results we get:
The first thing you might notice is that the coefficient on the “More Capable” variable is the wrong sign. This suggests that an increase in the “more capable” results for the Coalition actually leads to a decrease in the primary vote, the exact opposite of what was expected. However, the results are completely and utterly statistically insignificant (with the p-value= 0.86). Similarly, results of the “more capable” question don’t even explain 1%, or even 1/10th of a percent of the variation in the Coalition primary vote.
So what does all this mean?
Well, it means that the ALP vote can rise or fall with big movements in the “more capable” answers. But since the ALP results didn’t move from last time in todays Newspoll question (stuck at 29), and since the answers for the ALP are historically the same in both polling series, we can say that todays poll is completely irrelevant as far as the ALP is concerned.
As for the Coalition, their results increased today from the last poll result (increasing from 51 to 53), a movement within the margin of error and hence not worth talking about to begin with. But if one felt compelled to talk about the results, the statistical evidence tells us that the Newspoll measures of the economy ARE NOT key vote changing issues for the Coalition, the answers to these questions DO NOT have any relationship with the primary vote of the Coalition, and that, if one were to stretch the absolute limits of credibility to infer something from the data, the best that could be said is “Better economic management more likely hurts the Coalitions primary vote than helps it”.
More stable minds would simply say “there is no relationship – period”, which is about the same that can be said of The Oz’s polling interpretation and reality at times.
I’ve been a bit slack on the comments over the last 2 days – sorry about that folks, been flat out.