Wednesday, April 21, 2010

Lessons UK Pundits Can Learn from Canadian Elections

I am writing this because of the inability of people in the UK to see what could very well be happening in their election and suggest that Canada is a good example to look to understand what might happen on May 6th.

The UK election scene has been dramatically stable for several generations now, really the last election that held any surprise was the 1945 election when Labour defeated the wartime government of Winston Churchill.  It is because of these years and years of the same script that pundits in the UK seem to be unable to see how the election may play out.  That the swingometer used in the UK could be so certain and accurate always astonished me given my experience in Canada.

The US, Australia, Ireland and to some extent New Zealand, all have had similar fairly stable long term political trends in their elections, there have been very few surprises over the years.   They all reinforce the old thinking for the UK and how elections happen

Canada offers the best examples of extraordinary election results.   For PoliSci geeks and pundits we also have the advantage that we have a federal government and 10 provincial ones.  This means we have many more elections happening Westminster parliamentary election happening than anywhere else in a given four year period.

I will quickly go into several election results in the last generation to show what happened.   Then I will suggest to people a different way to think about elections and therefore the possible outcomes.  Skip to the bottom if you do not want to read the details of Canadian election results.

1990 Ontario Election
The Liberal premier of Ontario called an early election to get another mandate, but in the course of the election the public turned on him and the NDP won a majority government.  The NDP went from 19 of 130 seats to 74 out of 130 seats and Bob Rae was in government.   The increased their vote by close to 1/3 and swept seats they never ever dreamed of winning.

Numerous places where the party had finished a distant third in 1987 were won in 1990.  Wentworth North, the seat considered the safest by the governing Liberals was lost to the NDP.   In 1987 the Liberals won with 54% of the vote and the NDP was third with 22%.  No one on any planet could have predicted that loss.

The NDP had no idea or hope that they would win the election and were completely unprepared to be government.   One elected member was elected and so unprepared that he did not even own a suit to go to provincial parliament in.

1991 BC Election
In 1991 in BC the decades long battle between the NDP on the left and Social Credit on the right all fell apart at one moment during the leader's debate.

Sorry the quality is so bad.

The Liberal leader Gordon Wilson made one comment that won him the debate.   The Liberal party went from 6% of the vote and no seats in the previous election to 33% of the vote and 17 seats.   Some of the members elected by the Liberals were little more than paper candidates and did not think about getting elected.   One candidate that was elected found out when the media and losing member showed up at his door to congratulate him.

No pundit predicted this result

1993 Federal Election
Canada was seeing the end of an increasingly unpopular Conservative government, but no one expected the government to be utterly destroyed.   At the start of the election the Conservatives where behind in the polls, but not desperately so, but their numbers dropped through out the election.  At the end of the election the Conservatives went from 169 seats to 2.   The Prime Minister herself was defeated.

This election also saw the rise of two new parties from almost nothing to holding over 50 seats each in a parliament of 295 members.  One of the parties, Reform, was not expected to win more than a handful of seats, but finished with 52 seats.

2001 BC Election
The result on many levels was not that surprising, it was a huge landslide for the Liberals over the NDP, but the scale that the NDP vote fell to was unprecented.   The NDP went from being a majority government to only have two members elected.   The Liberals held the other 77 seats.

Pundits were certain the NDP would be trashed and lose badly, but the idea that the party would be almost completely shut out was inconceivable.

Other Extraordinary Results
Alberta 1971 - Out of nowhere, the 36 year Social Credit government is defeated by the Progressive Conservatives which have held a utterly dominant lock since then though look like they may lose the next election.
Manitoba 1988 - The governing NDP came third behind the Liberals which had a single seat before the election.   The Liberals faded away in future elections
New Brunswick 1987 - Liberals won over the governing Conservatives and took all 58 seats.
New Brunswick 1991 - Liberals win, but a new far right wing party suddenly won as the official opposition
Nova Scotia 1998 - The long time minor third place party, the NDP, finished tied in seats with the governing Liberals.  The NDP is now the majority government in the province.

What are Voters really doing?
This has become an important question to ask in Canada because of some of the dramatic elections we have seen.   Because the results defy the traditional predictions of pundits, there has been some serious though given to what is actually going on.    Can we figure out why these extraordinary results happen?  One person researching this is Werner Antweiler at UBC.

Werner is a professor in the Sauder School of Business at UBC and has been running research into using an online prediction market to estimate the results of elections.   He has run this as the UBC Election Stock Market.   The results have been close to the final results, especially with respect to seats won.   It has also often been ahead of the curve on the momentum in an election.

During the course of the research on election predictions he has identified a factor rarely considered in election outcomes, the non-voter in a past election intending to vote in this election.   His research shows that in in most elections there is a bigger swing from a party to non voting and vice versa than between parties.  When one applies this thinking to the extraordinary results in Canada, the model starts to come closer to the results.

  • In 1990 in BC there was a swing from Social Credit to Liberal, but many more Social Credit supporters stayed home and quite a number of non voters came out and voted for the Liberals.
  • In Ontario in 1991 many of the seats lost by the Liberals owed more to Liberal supporters staying home than a huge shift to the NDP.
  • In 2001 in BC the abysmal result for the NDP is directly from the majority of their supporters not voting in the election.

This factor has rarely been considered in elections and I think could be the major impact on this election in the UK.

If we use this lens for current UK election, what I believe is going on is that a lot of traditional Labour voters will be choosing not to vote.  I suspect there are some Conservative supporters in the same boat, but many fewer.   Their people stayed home in 1997.   I suspect some of the growth in the Liberal Democrat vote is from people that did not vote in the last election.

If you are not convinced that this happens in the UK, look at by-elections.  They are often good examples of surprising results.  I hold that the results happen in large part because of low turn out, especially from governing party supporters.

Choosing a random Labour held seat, Barnsley Central, and applying some of this thinking:

First using the way most UK pundits think a the moment, the result would be:
  • Labour - 14000
  • LibDem - 7200
  • Cons - 3800

Now we apply some of the shifts in voting/non voting from election to election.   Let us say that 1/4 of Labour supporters are not changing their vote, but staying home, another 1/6th choose to vote Liberal Democrat.   Meanwhile an extra 3,000 people vote that did not vote in 2005 and they support the Liberal Democrats.   Result:

  • Labour 10,000
  • LibDem 10,700
  • Cons 3800

It becomes winable for the Liberal Democrats.  Think about that, Labour won with a 12,732 vote majority, 61% of the of the vote, in 2005 and they could lose in this election.  Certainly there have been results like this in Canada, this happened in Calgary North in 1993.

You can apply this over and over again across the country and see a very dramatic shift in seats could very well happen.  Labour being reduced to less 100 seats is possible.


Aron said...

Interesting piece. I'm with you in being skeptical about the reliability of the 'swingometer', especially in a three horse race and volatile voter base.

Michael Maguire said...

As a Canadian now living in the U.K. I thought this was interesting and relevant. I find it odd that people here constantly dismiss the likelihood of the Liberal Democrats getting into power (although that's changing slowly after the first few debates).

The video of the BC debate reminded me of Nick Clegg's performance in the first tv debate here in the U.K.

Although I will be voting for Nick and the Liberal Democrats (who would like to reform the first-past-the-post system in the U.K. to be more proportional) I thought it was also a bit of an interesting rebuttal to the necessity for reforming the system, which as you've demonstrated still provides opportunities for radical change in elections.

Stephen Glenn said...

Very interesting. In fact I've noticed that some of the swingometers even those that are allowing three way adjustment are not allowing for some of the current Lib Dem swing as predicted in the opnion polls.

We live in interesting times. The next two weeks of hard graft are going to be very interesting.

Mark Greenan said...

GREAT blog. This actually should be required reading for UK pundits this month.

I also share the skepticism about uniform-swing predictions, particularly as concerns the Lib Dems.

Also, I found Prof. Antweiler's research fascinating. Do you know if he's published much from his election research.

Finally, Michael, in close to a decade in the proportional representation movement in Canada, I must say I yet to hear someone argue we need electoral reform because it "provides opportunities for radical change in elections".