The legend of the delegated convention

They’re entertaining, but not quite as wild and crazy as you remember
Rob Silver

This weekend’s Ontario Liberal leader election and the upcoming Quebec Liberal leadership election could be the last of a dying breed—the wild, unpredictable delegated convention. Fantastic drama, questionable democracy: such conventions are crack to political junkies.

Any smart political junkie knows that delegated leadership conventions have some steadfast parameters. A former Ontario cabinet minister who is a really smart guy (really, he is) summarized them recently as follows:

In delegate-based leadership contests, frontrunners are typically doomed to lose — except where there is an heir apparent, such as Chretien succeeding Turner, or Martin succeeding Chretien (and I suppose, Ignatieff succeeding Dion). The same held true for leadership contests in the other major parties — at least until they got rid of the old-school delegate convention format and replaced it with something akin to trolling for Twitter followers.

Front runners are doomed, unless there is an heir apparent. Gotcha.

He then set out three specific rules:

For delegate conventions, like the one coming up on January 26th, there are three rules of thumb. One: the frontrunners become a target for the other candidates — if they don’t win on the first ballot, they tend to lose on the last ballot.

Win on the first ballot or lose on the last. I’m following.

Two: the candidate who offends the fewest and befriends the most wins the race. All the leadership candidates liked Dalton McGuinty in 1996, and most delegates did too. Nobody expected him to win. But he was the most inoffensive, and built a coalition that brought him victory. The same was true of Lyn McLeod in 1992, and of Stephane Dion in 2006.

And finally.

Three: mainstream media coverage of a leadership race is unreliable, unless you follow the media with Rule #1 in mind. The media focus primarily on the frontrunner or frontrunners. In 1996, that was Gerard Kennedy and Joe Cordiano. Neither won. In 2006: Ignatieff and Bob Rae. Both lost.

There’s a well known phenomenon known as recency bias which is roughly the tendency to think that trends and patterns we observe in the recent past will continue in the future. The 1992, 1996 and 2006 leadership races are (relatively) fresh in our minds. All of those were won by candidates who “came out of nowhere.” The frontrunner in those races met all the “rules” set out of above. Thus, all delegated leadership conventions must follow these same rules, right?

Not even close.

I found 35 delegated leadership races in Ontario or at the federal level since 1958 held by the Liberals, PCs and NDP at both levels. That does not include races where there was only one candidate on the ballot (ie. the 35 does not include Michael Ignatieff in 2008 or Robert Nixon in 1967). While the rules of how delegates are selected have varied wildly over the years, for the purposes of this piece, I’m far more interested in the dynamics once delegates get to the convention—who is leading after the first ballot and how things unfold from there.

Of the 35 races I looked at, the candidate who placed first on the first ballot won 28 of 35 times (80%). Full stop. There have only been seven instances in the last 55 years in Ontario or at the federal level where somebody other than the frontrunner on the first ballot won. In other words, it has happened but it is not common and is certainly not a “rule.”

The average first ballot vote for the leader was 45.28%. That includes candidates who won on the first ballot often with massive vote totals. If you only count races that went more than one ballot—which happened in 25 of 35 races—the average first ballot vote was 36.4%. So frontrunners have won even when they didn’t come particularly close to getting over the finish line on the first ballot.

Where do frontrunners want to be on the first ballot in order to be “safe?” There’s no steadfast vote level that you need to cross. Of first ballot leaders who won on subsequent ballots, the lowest percentage was Robert Stanfield in 1967 at 23.3%. Only two other times did first-ballot leaders who ultimately won fail to get at least 31.5% of the first ballot vote (NDP leader Audrey McLaughlin in 1989 at 26.9% and OLP leader Andy Thompson who got 28% on the first ballot in 1964).

Again: the vast majority of frontrunners on the first ballot at delegated conventions have gone on to win. It is much better to be in first than not. Most of the time, they had more than 31.5% of the vote on that first ballot, but sometimes they didn’t.

On to those famous upsets. The seven first-ballot leaders who did not win are:

Walter Harris OLP, 1958, 39%
Kelso Roberts OPC 1961, 21%
Claude Wagner PC, 1976, 22.5%
Joe Clark PC, 1983, 36.5%
Murray Elston OLP, 1992, 30%
Gerard Kennedy OLP 1996, 30%
Michael Ignatieff LPC, 2006, 29.3%

Those seven averaged 29.6% of the vote on the first ballot with a pretty wide range of vote. So how did these seven “upsets” happen?
In four of the seven races, the eventual winner finished second on the first ballot.

1958 OLP:
Walter Harris 39%
John Wintermeyer 34%

1961 OPC:
Kelso Roberts 21%
John Robarts 20%

1983 PC:
Joe Clark 36.5%
Brian Mulroney 29.2%

1992 OLP:
Murray Elston 30%
Lyn McLeod 27%

In the four instances when a second-place first-ballot finisher won, they came back from 1, 3, 5 and 7.3 points behind. It’s hard to call any of these “out of nowhere upsets.” So we’re down to the three instances where there were true “out of nowhere” results. Most of us are quite familiar with those conventions and what happened and the purpose of this post is certainly not to dissect how those upsets unfolded. But a few quick facts:

Michael Ignatieff in 2006 blew the biggest lead of any frontrunner—nine points over Bob Rae (ie. every frontrunner who has led by 10 points or more on the first ballot, has won). Dalton McGuinty won from the furthest back (12 points behind the leader, he was also the only winner who was fourth on the first ballot). Joe Clark had the lowest first-ballot score (by a wide margin) of any eventual winner at 11.7%. And Dalton McGuinty and Stephane Dion are the only other winners to get less than 20% on the first ballot.

So to summarise: 28 of 35 delegated conventions were won by a candidate who finished first on the first ballot, 32 of 35 were won by the first- or second-place candidate and 32 of 35 times the winner had more than 20% of the vote.

Upsets can, of course, happen. Who knows how this weekend will play out. Just understand that “conventional wisdom” that delegated conventions are wild affairs where frontrunners are doomed unless they win right away is based on recent, short-term memory and not supported by history.