Queensland 2015: A Case of Regression
How spectacular was Labor’s win, and how devastating a defeat for the LNP? Statistics may have the answer.
I’m not going to discuss Freud’s psychoanalytic theories or go down rabbit holes of anyone’s psyche in this post. I’m reflecting on the outcome of the Queensland election and offering an opinion on just how miraculous it was for the Australian Labor Party (Labor) or disastrous it was for the Liberal National Party (LNP) — from a statistical viewpoint.
Queensland’s Labor formed minority government on February 13, 2015 after winning 44 of the 89 seats in the election held on January 31. Former Premier, Campbell Newman, hailed as the LNP messiah in the 2012 election, when Labor was reduced to just 7 seats, lost his own seat this time around. Between the two elections, Labor picked up another 2 seats through by-elections bringing their total to 9 seats.
So when the numbers were being crunched on election night January 31 and the morning after, ABC online led with the headline “Queensland Election: Massive swing to Labor as LNP routed”. While this headline is factually true, it does leave the reader with the impression that Labor won a stupendous “against all odds” victory, and the LNP suffered a crushing defeat.
So just how spectacular was Labor’s 2015 win, and how devastating a defeat was it for the LNP? Perhaps, it’s time to look at historical patterns.
Between 1932 and 2015, the number of seats in Queensland’s increased from 62 to 89 in several steps. To get around the changing base figure, I’ve computed each major party’s percentage of the total number of seats at each election, and this is visually represented in Chart 1 and Chart 2.
Chart 1. Percentage of seats won by Liberal National 1932–2015
Chart 2. Percentage of seats won by Labor 1932–2015
I’ve computed the mean (average) and standard deviation of the percentage of seats for each party between 1932 and 2015. The standard deviation is a measure of how tightly the data points cluster around to the mean. I’ll expand on the significance of the standard deviation a little later. The black line running across the width of each chart is the mean number of seats for that party. The two green lines running across the width of each chart are +/- 2 standard deviations from their respective means. These green lines are indicators of how extreme a data point is.
In the 30 elections from 1932, the only time when the LNP’s percentage of seats departed from their mean of 45.5 by more than + or - 2 standard deviations was 2012. Labor experienced a similar swing away from their mean (48.9), albeit in the opposite direction. For both parties, the percentage of seats held in the election immediately before (2009) and immediately after (2015) were much closer to their respective means.
There is only one occasion in 30 elections when the swing for Labor and Liberal-National coalition (in Queensland, the Liberal Party and National Party merged in 2008 to form the Liberal National Party) was just shy of 2 standard deviations from their respective means and that was 1974. Once again we see the same before-after pattern.
There is a statistical term that describes this pattern of an infrequent extreme departure from the mean only to be followed by outcomes that are closer to the mean — regression to the mean.
Regression to the mean is common in many aspects of life. The nobel-prize winning psychologist, Daniel Kahneman, tells a story of trying to disabuse a flight instructor of the notion that praise was bad and criticism was good for cadet performance. The instructor pointed out that after he had praised a cadet for performing a clean maneuver, the trainee performed less well subsequently. The instructor also noticed the opposite was true when he criticized a student for a weak performance.
Professor Kahneman pointed out that these were simply cases of regression to the mean. Although, a pilot’s performance is based on considerable skill, performance will vary randomly with each maneuver. So when a cadet executes an extremely clean maneuver, they would have had a little bit of luck in addition to their considerable skill. On the next maneuver, the contribution from luck would probably disappear and their performance is negatively impacted after the praise and not because of it. The converse is true for improved performance after chastisement.
Every major political party wants to form government, so winning is celebrated and losing is commiserated over. But the data since 1932 strongly suggests that Queensland 2015 was simply a case of the Queensland voting public returning to their normal voting pattern after a marked departure in 2012.
Notes:
Liberal-National mean: 45.5 percentage points
Liberal-National standard deviation: 19.6 percentage points
Labor mean: 48.9 percentage points
Labor standard deviation: 18.1 percentage points
Data source: http://en.wikipedia.org/wiki/Legislative_Assembly_of_Queensland
ABC on Queensland 2015: http://www.abc.net.au/news/2015-02-01/queensland-election-massive-swing-to-labor-as-lnp/6060330
Thanks for reading. If you enjoyed this article or found it useful, please scroll down and click the “Recommend” button. It’d mean a lot to me.
Follow
Twitter: @kelvinparam
Google+: +Kelvin Param
Facebook: https://www.facebook.com/lifemeetsdata
