Close races: really not important?

This Monday, the Hill Time online was showing how very few ridings are consistently what we call a close race, i.e: where the winner has a lead of less than 5% over the opponent(s). This is interesting but I do feel it's only half the story.

Of course, it makes sense that a close race last election won't necessarily be one this time around. The main reason? Parties' levels of support change between election. So if a riding was a close race in 2006 in Ontario between the Conservatives and the Liberals when those parties were at respectively 35.1% and 39.9%, it's only logical that the race is over when the same two parties now stand at 39.2% and 33.8%.

Nevertheless, I do think the Hill Time is too quick on dismissing the importance of being a close second during an election. I think what we should really be looking at is whether those close races changed colour the next election. So here are some table for the 2004 to 2008 elections.

2004 close races close win kept in 2006 close in 04, win in 06 net diff.

CPC 36 25 19 8 2
LPC 51 23 5 3 -15
NDP 22 7 7 7 7
Green 0 0 - - -
Bloc 8 2 2 6 6

2006 close races close win kept in 2008 close in 06, win in 08 net diff

CPC 33 18 17 11 10
LPC 44 22 7 3 -12
NDP 13 6 4 5 3
Green 0 0 - - -
Bloc 9 3 1 1 -1

2008 close races close win

CPC 31 17

LPC 29 12

NDP 12 6

Green 0 0

Bloc 10 5

A couple of interesting things here. First of all, notice how inefficient the NDP votes was back in 2004. High enough to put this party in the race in 22 ridings, but only 7 wins? That's a really low rate of convertion. It seems that over the year, the NDP has become much more efficient and that alone could explain why the NDP can actually lose some percentages but actually win more seats (kinda like in BC or Ontario).

Second of all, the Liberals just got destroyed the last two elections. Their losses in previously hot races are huge. By simply looking at those losses, we cna explain 27 out of the 58 seats lost between 2004 and 2008. If you ask me, this is significant. It shows that a lot of the action is actually happening in those close races.

That can also help explaing the rise of the Conservatives. Let's just look at 2008. Out of the 33 ridings where the Conservatives were in the race in 2006, they won 28 of them in 2008! On the other hand, for the Liberals, the numbers are 10 for 44. Don't tell me this didn't have a massive impact of the victory of the Tories.

So to sum up, I agree that being a close race in one election is not necessarily a good predictor of whether this riding will be a hot race next time. But those close races are important and should be watched closely because they are the sources of a lot of the action/changes. Moreover, looking at how the Conservatives are slighlty but surely grabbing more of those previously close races, I wouldn't be surprised if they actually manage to get a majority the next time around. I mean, look at the numbers for 2008. They lost 14 ridings by a margin of less than 5%. With a good riding targeting, this party won't even need to increase its level of support substantially to win this majority. But this is another story.