Monday, August 05, 2013

The Ontario By-elections: a strong result for the PCs

The title of this post is at odds with the received wisdom about the by-elections. That is simply because the five by-elections went 2 Liberal, 2 NDP and 1 PC.  All five had previously been held by the Liberals but the PCs thought they could win as many as four of the five so gaining just one seemed a disappointment.

PC Leader Tim Hudak
However, when we look closely at the voting and I apply my reverse vote estimation model we get a different story.  The model essentially takes a single by-election and tells us what it would mean if blown up (by a mathematical conversion) to apply to the province as a whole. While the results should be taken with a grain of salt, I think they do give us some insight into current trends.

In the case of every constituency, the projection from this model would mean a PC majority: a Tim Hudak government.

If we do a simple average of the province-wide vote estimate including all five ridings, we get the PCs with a huge lead in vote share:

Liberal PC NDP GP Other
Average Vote Projection 22.4% 40.9% 29.7% 3.9% 3.2%

This would give us the following outcome in terms of seats:

Liberal PC NDP GP Other
Average Seat Projection 12 65 30 0 0

The PC vote share average is driven up by an exceptionally strong showing in the riding they won: Etobicoke-Lakeshore.  However, in every other riding their result is quite consistent with the 34 to 36 per cent they have been getting in provincial polls. The projected PC majority is clearly a product of the vote splitting between the Liberals and the NDP and has echoes of the 2011 federal results in Ontario when the Harper Conservatives won over 70 seats in Ontario.

Here is a table of the overall vote shifts from 2011:

Party Shifts from 2011
August 1 2013 Byelections
ConstituencyLiberalPCN.D.P.GPOther
Etobicoke—Lakeshore-8.3%18.3%-9.3%-0.4%-0.3%
London West-29.8%3.3%20.2%1.8%4.5%
Ottawa South-6.5%5.2%0.9%-0.1%0.5%
Scarborough—Guildwood-13.1%2.1%9.0%0.9%1.2%
Windsor—Tecumseh-30.9%-0.7%28.5%1.4%1.7%
Total-17.3%6.3%8.7%0.7%1.6%

The by-election results were actually very bad for the Liberals, despite their two victories.  These were Liberal held seats, and most had voted Liberal by substantial margins in 2011. The Liberal Party lost ground everywhere. In two seats, London West and Windsor Tecumseh, there was clearly a huge shift from Liberal to NDP, while the PC vote share remained relatively stable. The NDP also appears to have gained significantly from the Liberals in Scarborough-Guildwood.

Note that there was the appearance of a large shift in vote share from both the NDP and Liberals to the PCs in Etobicoke-Lakeshore where Doug Holyday, a long-time popular Etobicoke municipal politician, captured the seat for the PCs.  However, one can never know for sure in any election whether there could be a double shift happening where, for example, some NDP voters switch to the Liberals at the same time as Liberals are switching to the PCs. In terms of vote share Etobicoke-Lakeshore was the best Liberal result, and their second smallest vote loss, so it seems at least a possibility that something like that happened here: a polarization that led to NDP losses to the Liberals, partially offsetting significant Liberal to PC vote shifting.

Of the five seats it appears that two, Etobicoke-Lakeshore and London West, where New Democrat Peggy Sattler vaulted the NDP from third place in 2011 to first, strong local candidates played a critical role in determining outcomes (one reason why the vote projections above should not be taken too literally). Windsor-Tecumseh was a big win for the NDP.  It continued a long-term trend in the motor city from Liberal to NDP.  In Ottawa South and Scarborough Guildwood, despite winning, the Liberal party registered a loss of vote share from 2011.

The by-elections also continued the trend of shaky polling results.  This 308 blog post sums up the situation well.

The by-elections may predict a change in the next provincial election but they are not a completely reliable indicator.  In October 1978 there were fifteen by-elections across Canada following a decision by Pierre Trudeau not to go to the polls (among the victors in those by-elections was a young Bob Rae). Taken at face value the overwhelming message coming from the 1978 results suggested an impending large PC majority. Instead in 1979 the Joe Clark PCs, while victorious, secured only a minority government. Among the PC winners in 1978, five would go on to lose to Liberals in 1979.

The Kathleen Wynne Liberals should be considered down but not out.

Saturday, August 03, 2013

The Green Vote and Strategic Voting in the 2013 BC Election

The May 14, 2013 BC Election created an unusual circumstance that allows us to make an assessment of the potential of strategic voting.  The circumstance was the universal perception that the NDP was going to win - the consequence of the inaccurate polling and punditry (see TC's three previous posts). This post will analyze the Green vote in this election and look at the strategic implications.

Even if Green voters thought that the NDP was preferable to continued governance by the Liberals (a majority of New Democrats and a plurality of Green supporters made the other party their second choice in this poll) this unique context meant Greens had no reason not to support their first choice given the expectation that the NDP would win without any help. There was nothing to hold back Green performance in this election with the exception that the party did not run a full slate of candidates, something they did do in 2009.

On its face the Green performance in BC was little different in 2013 from 2009.  The Green share of the popular vote actually declined slightly from 8.21 % in 2009 to 8.13 % in 2013. Nevertheless, the Greens and most observers saw the election result as a triumph because they elected their first member to the legislature, Andrew Weaver in Oak Bay-Gordon Head.

Andrew Weaver - Oak Bay-Gordon Head

In fact, the Greens performed better in the popular vote than might seem to be the case at first glance. If we look only at the Green share of the vote in the ridings they did contest, we find the party capturing 11.2 % of the popular vote, a significant improvement from 2009, when the Greens won 8.5% of the vote in the same list of constituencies. This showing is second only to the 12.4% of the popular vote overall the Greens garnered in the 2001 election in the context of collapsing support for the NDP.

The question arises then: what impact did the absence of strategic voting on the part of Green voters potentially have on the election outcome.  I ran a number of scenarios to analyze this. Scenario one: if we absolutely take away all Green votes and give them to the NDP while leaving Conservative votes in place the NDP would have won 46 seats to the Liberals 38 (the actual result was Liberal 49/ NDP 35). This suggests a purely hypothetical maximum for the NDP in 2013 but is not realistic.

A more realistic way of looking at the potential of strategic voting is Scenario two: take one third of Green votes and deliver them to the NDP, as it is highly likely that only a minority of partisans would vote strategically.  In this scenario the Liberals would have led the NDP 45 seats to 39 compared to the actual (in this scenario the Green seat goes NDP). If we add as a parallel shift one third of Conservative votes to the Liberals the margin increases to 47-37.  These are hypothetical projections, but it is clear that Green voting combined with the flawed polling did not cost the NDP the election. The failures of their campaign accomplished that.

There was some evidence of strategic voting by BC Conservative voters.  Early in the campaign, the party was at 11 to 13 % in the polls (without forgetting their doubtful accuracy) and had been over 20% during part of 2012.  The attack ads aimed at the NDP by the BC Liberals no doubt had an impact in bringing these voters home (most were previous BC Liberal voters). The Conservatives ended up with just 4.8% overall (they averaged 7.3% in ridings where they had a candidate).

Strategic voting potential, however, was there. TC heard anecdotal evidence of a willingness on the part of some Green supporters to switch to the NDP. Would the Greens have finished closer to 8% if there was a context to motivate strategic voting? The question will remain unanswered.

Perhaps better described as tactical voting, this type of second choice voting is a fact of life in Canada and likely to be of increasing consequence in future elections, providing polling companies can do a better job of measuring vote intentions.

SMALL ADDENDUM: One observation I made in looking at Green voting is that in general Green and NDP support overlap geographically: both parties are strongest on the coast, Vancouver Island and the lower mainland.

In my next post, I will comment on the Ontario by-elections. I have come to some conclusions about the results, some of which may seem counter intuitive.