Polling from 2008 to 2012
March 27, 2012 5 Comments
Two polling firms (Ipsos and Think HQ) are reporting that the PCs and Wildrose are in dead heat. That alone is much fodder for discussion, but I soon asked myself, “What was the road to 2012 like, in terms of polls?”
To answer that, I went to Election Almanac and plotted the polling data they have archived since the 2008 Alberta election. The chart is below.
Before I continue, I want to emphasize that causality is impossible to determine. All this chart shows are trends, and when the trends are compared against events that occurred during this time period, arguments can be made for correlations between trends and events. I have marked major events that occurred for each party. There are too many other major events (like the spring 2011 health care crisis or the recent string of government scandals) to include on this graph, but the political junkies looking at the graph are bound to have a list floating in their memory bank that they can consult as they look at the numbers.
The methodology for constructing this chart is simple. All data is from Election Almanac. (As an aside, I will be tracking the polls this election and will post updates on my blog.) The source and date of each poll is noted on the chart. For months where there were multiple polls from different companies, the mean is calculated across all polling firms for each respective party.
An interesting data point on this chart is where the Wildrose were reported by Angus-Reid to have surpassed the PCs in December 2009 and January 2010. This was the height of the post-Danielle Smith honeymoon, which was the same time that Rob Anderson and Heather Forsyth crossed the floor from the PCs. I have not found Angus-Reid’s tables on that poll, but I do remember that, at the time, I doubted the accuracy of those numbers. Not only was that the only time in the inter-election period that the Wildrose was above the PCs, but a gap of almost 15 points seems implausible. Moreover, a difference of less than 10 points between the PCs and the Liberals also seems implausible. However, without the tables in front of me, I will not saying anything beyond, “The numbers seem curious.”
March 2012 is a completely different story. Three companies (Abacus, Ipsos and Think HQ) all reported similar numbers. Abacus had the margin at five points. Think HQ and Ipsos, which were conducted at the same time and later than Abacus’ poll, measured the gap at two points and nil, respectively. Think HQ also asked questions about momentum (see the linked PDF), the responses to which further corroborate the trend. Thus, I am inclined to believe that the PCs and Wildrose really are that close. Of course, what happens 27 days from now is completely different story, but for now, the two parties are neck and neck. In fact, they are so close that the PCs actually dug up a decade-old editorial written by Danielle Smith, which is arguably designed to drive a wedge between the libertarians and social conservatives in the Wildrose in an effort to chip away at their support. The Wildrose, on the other hand, can almost smell victory, so they have done a good job sticking to broad and universally-appealing messaging in a bid to broaden their support beyond their base.
Another interesting feature is the competition that occurs on either side of the political spectrum. PC and Wildrose support are inversely correlated, suggesting that one’s gain comes at the other’s expense. The same thing can be seen on the left between the Liberals and NDP and, later, the Alberta Party. There were a few periods when support for all opposition parties were correlated (Dec/08-Jun/09, May/11-Nov/11), but these are the exceptions rather than the norm. The latter period is possibly due to the PC leadership. Parties occupy ranges on the political spectrum, and the PCs’ right flank (think of Ted Morton) overlaps with the Wildrose and the PCs’ left flank (Alison Redford) overlaps with the Liberals. Moreover, that the Wildrose rebounded after the leadership, but the parties on the centre and left did not, suggests that many moderates and progressives did, in fact, become ‘Redford Tories’, and that more right-leaning PCs moved to the Wildrose. However, as I mentioned earlier, only individual-level data would be able to scientifically validate that theory, so we will have to wait until after the election when political scientists conduct a post-election study.
I will leave my comments at that. Feel free to add your own comments or to share this graph around. I hope to put together a similar graph for the polls that come out during the election.