Thursday, April 14, 2011

Examining EKOS Canadian Federal Voter Intent 13 April Data:: Conservatives About to Slide?

EKOS polling data of federalist parties for 2011 w/selected regions. Poll taken 11-12 April 2011.
Click for larger view
The latest EKOS poll [pdf] [house effects on 308] was done just before the debates. It should worry the Conservatives, despite the party's overall strong showing. The Liberals have narrowed the Ontario gap, they're losing ground in BC {along with an NDP surge}, and the Liberals have rebounded in Atlantic Canada. I think this is good news for the NDP, particularly in BC and Québec. I'm sure they wished that the numbers in Ontario were stronger, which would cause sleepless nights for the Liberals.

Regarding the debates, I watched and Tweeted on both, although my French isn't good enough to watch the non-dubbed feed. I think Impolitical had a good overview of the English debate. My take on both is that this was Harper's big pitch for a majority. While the pundits may say he looked "prime ministerial", I thought he was playing a risk-averse game that was short on details or ideas. I can't say I blame him. He has some pretty crappy cards to play, but he and his party chose those cards. Ignatieff may not have "hit a home run", according to the pundits, but he didn't need to. He needed to come across as credible, which he did, and the public, save for those who watch CPAC, finally got to see him in action. I think he might come across like an internationalist, but both the Liberals and the NDP have platforms that actually address the issues that concern Canadians, with the top issue being healthcare. 

Harper is playing a defensive strategy, but he wants to make a play for a majority. In marketing parlance, he wants to increase market share, but he's not doing a good job of differentiating himself and the Conservatives. He's gone mainstream and away from the old Reform wing, but playing to Reformers in the West, the moderates, the suburbans, and to new Canadians is an amalgam which gives his party few degrees of freedom in the current economic climate. So, what's the sell? Outside of the Conservative base, Harper really doesn't have much to offer, given the budget and his defensive strategy. This allows the Liberals and the NDP to stake out territory that promotes ideas, but it remains to be seen if the public will start to turn their backs on the Conservatives.

Second Choices

Examining the second choice data, a curious pattern is emerging. The Conservatives are picking up second choice support by those intending on voting for the Bloc, NDP, and Liberals, but the mutual support of the Liberals and NDP is strengthening. The NDP and the Liberals are sharing more ideological space. I put the second choice data into network visualization software {Netdraw 2.081}::

EKOS Second Choices
Nodes represent federal parties. Inbound links (figures closest to each node) represent percent support of voters who support another party for first choice who support the focal party. Example: On the line between the Liberals (LPC) and the NDP, the 43.2 represents the number of Liberal supporters who consider the NDP a second choice. The thickness of the lines represents the second choice relationships between the parties. Recursive loops represent "no second choice" for a party. 
I decided to look at the same diagram, but looking at the percentages controlled for first choice support. In other words, each percentage on each line would be for the entire sample, not based on the degree of support for the parties. So, all the percentages on the lines {to and from each party} would add to 100%. This gives a clearer sense of how many votes can shift::

EKOS Second Choices, Normed by First-Choice Party Support
Nodes represent federal parties. Inbound links (figures closest to each node) represent percent support of all voters who support another party for first choice who support the focal party. Example: On the line between the Liberals (LPC) and the CPC, the 6.0 represents the percentage of all voters who are Conservative supporters who consider the Liberals as a second choice. The thickness of the lines represents the second choice relationships between the parties. Recursive loops represent "no second choice" for a party.

This diagram makes it clear how much ideological space the NDP and Liberals are sharing, representing over 20% of the electorate. The Conservatives are sharing around 10% with the NDP and the Liberals, but more Conservatives are supporting the NDP and Liberals than the other way around. The relationship between the NDP-Greens, Greens-Conservatives, and NDP-Bloc all represent 5% of the electorate, while the Liberal-Greens relationship is 6%. This reinforces the EKOS statements a while back stating that the Conservatives have little room to grow. A majority would require a much more aggressive campaign. The "bubble" campaign and Harper's performance in the debates shows a defensive strategy.

If Conservatives start slipping, ceteris paribus, the NDP and Liberals should pick up Conservative support at relatively similar levels. Interestingly, the NDP has the most potential to pick up support from other parties {adding the inbound links of 2.4, 2.9, 1.5, 12.4, and 7.1=26.3}, given the second choice pattern. Inbound links for the Liberals are 16.7 and 9.5 for the Conservatives. Nevertheless, I see this as good news for the Liberals and somewhat bad news for the Conservatives. The Liberals can pick up support without relying on the NDP slipping. The Conservatives' need to appeal to Grits and Dippers to up their numbers, which might be a tough sell with a defensive strategy. I think the NDP could take advantage here, but I'm not sure where. Most likely BC. Without regional second choice  breakdowns, further conjecture would be pushing the limits of good taste. 

The Centre of the Canadian Political Zeitgeist
Finally, I've been interested in looking at second choice data to see if there are shifts in electorate with respect to the patterns of party support. I've been using social network analysis {UCInet v6.198} to assess this by looking at the core-periphery statistic of multiplicative coreness. If we think of party support as blobs within the political zeitgeist, coreness attempts to determine which blobs are central to Canadian consciousness and which ones are on the periphery. The closer a coreness score is to 1 means a party is increasingly correlated with a core preference pattern. The closer to 0 means a party is on the periphery. The analysis follows::

The last analyses with 6 April EKOS data had the Conservatives with the highest coreness score at .768 with the Liberals at .458 and the NDP at .386. The Bloc was at the periphery with low national support and weaker ties to other parties and a score of .099. The Liberals and NDP are closing the gap. If the Liberal, Conservative, and NDP coreness scores start to get really close, this is really bad news for the Conservatives. The Conservative party would be sharing the core with parties they have alienated under Harper, which may have consequences in the near term of the election, as well as down the road.

Twitterversion:: [blog] Social network statistical analyses of EKOS pre-debate polls. #CPC's race to lose, but evidence of weaknesses @Prof_K

No comments: