There is much understandable handwringing over the recent British Columbia provincial election, which resulted in a shocking upset. The Liberals, long forecasted to be politically dead, won government with a strong majority. What is even stranger about the result is that for months the NDP had a strong, stable lead.
How do we explain this sudden change in fortunes?
Explanation 1 – Polling is broken
Eric Grénier of 308 Blog argued that there is no reason this should have been called incorrectly by the polling industry. There have been some other recent upsets, such as the Quebec provincial election and (especially) the Alberta provincial election. However, Grénier says there is no excuse in the B.C. case. Pollsters were trying to model for parties that previously never existed. Both the CAQ and Wildrose were new and therefore projecting their support proved difficult. In British Columbia the Liberals and NDP have been fighting it out for decades, and the 2013 election does not look so different from the 2009 election.
There is some concern about the methodologies being used. Internet panels seem increasingly suspect in the face of these recent elections. The tried and tested method of calling with live questioners seems to be the most effective way of measuring the electorate.
Explanation 2 – Low turnout
Low turnout does strange things to elections. Parties with the strongest, most motivated base tend to do well in these scenarios. It appears that NDP support was soft in British Columbia leading up to election day. Who knows how many people simply stayed home rather than vote? Low participation can open space for upsets. All of a sudden the third place party that consistently pulls in 15,000 votes has enough to win the election in a particular riding. This phenomenon is best seen in by-elections. Look at the most recent ones in Calgary Centre and Victoria. The Greens performed extraordinarily well, far beyond their usual number, partially due to low turnout.
Explanation 3 – Parties matter
I believe the best explanation for how the B.C. Liberals did so well is found in the same explanation for why the Quebec Liberals did so well. In both cases these centre-right parties had strong, clear bases of support and the mechanics to turn out the vote. With strong support among identifiable groups these parties are able to mobilize much more easily and win the election riding by riding. Volunteers calling up supporters, driving people to the polls, and getting their voters to the polls gives them a distinct edge. Get out the vote campaigns is critical in every election, and it is arguably how every tight election is one. To win parties must push voters to the polls and not merely hope they choose to do their civic duty.
Explanation 4 – Campaigns matter
Going into the last provincial election in Ontario it looks like we would soon be governed by Premier Hudak. As the campaign went on the Progressive Conservatives bled support as voters learned more about their platform, Hudak stumbled over a few key issues, and the Liberals mounted attacks. Very few elections in Canada have ended where they began. Campaigns matter or we would not have them. People become convinced, people change their minds, people choose to vote. The post-mortem from British Columbia seems to suggest that the NDP ran a poor campaign. The focus on the positive and not responding properly to Liberal attacks merely left them weak and vulnerable.
I think the four above explanations can help us watch the next election; don’t worry about polls so much, pay attention to turnout/engagement, party strength is critical and the course of a campaign is critical. All of these things were equally true before the B.C. election, but we tend to just accept the simplicity and clarity of polls and ignore everything else.