Thursday, April 26, 2012

Wildrose Voters got Cold Feet

In the days after Alberta's engaging general election for the 28th legislature, everyone is talking about the polls - specifically how wrong they were. And while they did not accurately predict the outcome of the election, it is hard to suggest that the methodology was wrong. The polls, regardless of method employed, were pretty consistent with each other - particularly in the last week. If you have a large number of samples that are reinforcing each other, it is likely that the have a good sense of the question being asked and people's honest opinions on the matter. Let's take a closer look at the numbers.

Most of the polls were reporting percentages based on decided voters, including voters who indicate which way they are leading. A few of the polls reported on the number of undecided voters and where reported that number was somewhere between 16 and 30 per cent. For sake of analysis, let's make a few assumptions and look at the numbers. Assumptions:
  • There are 2,265,000 eligible voters in Alberta.
  • 20% of polled voters were undecided (5 polls that reported undecided percentage reported: 16, 17, 18, 24 and 30% undecided).
  • Polls in the last week showed around the following percentages for decided voters: 41% (WRP), 33% (PC), 11% (Lib), 11% (NDP).
2012 Projected Voters 742920 597960 199320 199320
2012 Actual Voters 442429 567060 127645 126752
Retention 60% 95% 64% 64%

If you assume that people generally don't lie to pollsters, you have to assume that something happened between the pollsters phone call at home and the poll booth. While the PCs retained 95% of voters who told pollsters they would vote PC, the Wildrose only retained 60%.

There is an old adage that parties don't get elected to office as much as other parties are thrown out of office. This would be very appropriate for Alberta. In order to change government, you typically need a high voter turnout. In 1921 the United Farmers were elected with 75% voter turnout; in 1935, when the Social Credit was swept into office, turnout was 82%; In 1971, when the PCs were swept in, turnout was 72%. Estimates for Monday's election are putting turnout at around 57%. In order to toss out the governing Conservatives, the Wildrose would need to get about 40% of votes with a turnout of over 70% - they would need about 634,000 votes. Enough people told pollsters they would vote Wildrose, but not enough did.
Essentially what happened is a big chunk of potential Wildrose voters got cold feet.

Actually, the wedding analogy is pretty apt here. For the most part if a wedding leads to a successful marriage it is because the couple knows each other well, they have dated for a while, they were likely engaged for a while and as time went on the commitment towards getting married grows and solidifies. Alternately, sometimes a nice conservative man meets a young attractive energetic young lady, they quickly fall in love, think the world of each other, get engaged and 28 days later when the wedding is about to start, and the man has had a chance to learn a bit more about his lover, he wonders whether he's making the right choice.
Alberta was tying on the patent leather shoes and figuring out how to tie up the bow tie when it said to itself, "Can we make this work if she doesn't believe in climate change? What else don't I know about her?"

There is no doubt that fear led to the collapse: A number of people who said they would vote Wildrose stayed home, a number went back to date the PCs a bit longer. Also a number of people who said they would vote Liberal or NDP ended up at the ballot box switching to the PCs. Alternately, PC supporters were more motivated to get to the polls, they feared the chance that they might lose grip on the province (especially to a radical WRP) and they got out to vote.

The WRP only retained 60% of voters who told pollsters they would vote WRP. There are three main reasons for this (most of which can be described as cold feet): anger with the PCs was not as big as expected and people stayed home; large poll numbers made people feel complacent and they stayed home; and people who were less engaged (less likely to vote) were inclined to tell pollsters they would vote WRP because of the early hype. This drop of 300,000 WRP supporters was the biggest factor in the PC win.

Another interesting trend is apparent when the bulk numbers from 2012 are compared to 2008:

2008 Votes 64407 501063 251158 80578
2012 Votes 442429 567060 127645 126752
Difference +378022 +65997 -123513 +46174

From this analysis, it is fascinating to see how every party was able to gain supporters with the exception of the Liberal party. While overall participation grew by 366,000 voters, It would be silly to suggest that WRP support came entirely from new voters. While some new voters would vote WRP, they gained most of their votes at the PCs expense. While the PCs lost a chunk of voters to the WRP, they would have picked up a lot of new voters and a lot of voters from the Liberals. Finally, the NDP base is relatively stable (in Alberta they are used to losing ridings and don't mind voting NDP anyway). They would have gained some voters in their stronger ridings from the Liberals and they would have attracted some new voters. The growth in raw vote support suggests that the WRP, PC and NDP can all claim some victory in this election.

There is a bit of data here and a lot of speculation, so I would love to hear your comments.

No comments: