Monday, November 06, 2006

The problem of numbers

So I've just spent a little time playing around with the poll data from EKOS, and I've got to say it's fascinating reading.

The problem is that any meaningful discussions need to be drawn from small sub-divisions of the poll sample, meaning that the margins of error are quite large, larger in some cases than the hypothetical margins of victory. Secondly, in most of these divisions the number of people who are undecided is larger than any other group. So if we're going to draw many conclusions, we need to assume that the undecideds break more or less along the lines of the already-decideds.

With both of those provisos - and I acknowledge to potential for error is large, here - I think Rae is in a very good position right about now. Let me explain.

Basically, I took the EKOS numbers for the four also-rans, and tried to extrapolate where their support would go after the first ballot. What I came up with was basically that the vast majority of support would go to Rae and Dion, with less going to Kennedy and Ignatieff. By the time the four leaders are left on the ballot, you end up with something like:

Ignatieff 34%
Rae 25%
Dion 21%
Kennedy 20%

Again, remember the large margins of error involved here. So far, we don't know anything we didn't already know - R, D, and K are all in contention for 2nd, with Rae running strongest. The big question is who drops out first - Dion or Kennedy?

I'm assuming, just for this exercise, that Kennedy drops out and his supporters divide themselves according to EKOS' numbers. Like I said before though, Kennedy's supporters are very loyal, so it's possible he could put Dion or Rae in first. Let's assume though that he doesn't. That gets us an Ignatieff/Rae/Dion split of about 37/33/30. Ignatieff's support has grown almost none at all, while Rae and Dion have both made large gains.

Dion, however, hasn't made enough gains to go on to the next ballot, and this is where Ignatieff's lack of growth really kills him. Almost nobody lists Ignatieff as their third choice, in the EKOS polls. 90% of Dion's supporters choose someone other than Ignatieff, and Ignatieff tops the list of people the delegates DO NOT want to see become leader. So if Dion drops out in third, I expect the vast majority of his support - at least 2/3, as much as 90% - to go to Rae. And that's enough to put Rae over the top.

Of course, there's still a month to go before the convention, and there's plenty of stuff that can still happen. In particular, Kennedy could throw his support behind Dion, and Dion could get surprise support from people who'd like to see a Liberal who's belonged to the party for more than 1 year win. (Rae comes in 2nd on the DO NOT elect list, behind Ignatieff.) More interestingly, Ignatieff could start to lose support if he doesn't take it after the 2nd ballot.

Politics, of course, is not algebra. We can't make predictions based on fuzzy math. Nevertheless, the fundamentals are against Kennedy at this point, and Dion has a very hard road ahead of him. Based on the data available today, I'd bet on Rae or Dion, and against Ignatieff or Kennedy. We'll see.


Charlie Barnard said...

The only thing this poll tells us for sure is that Iggy is unlikely to win. If there are no major changes by the convention then Iggy supporters would be wise to drop him on the 2nd ballot and vote for their second choice.

UWHabs said...

Everyone does all this math, but there's the page listing the "Rae vs. Iggy" comparison and it's 41-42-17, a virtual tie.

I think what will happen is Kennedy or Dion will drop out and endorse the other, who will pass Rae, and set us up for the final against Iggy.

Charlie Barnard said...

A more accurate way of putting things is that Ignatieff won't win unless one of the other top 4 candidates endorses him. Bleed will not be enough. So who out of the top 4 would endorse Iggy? Surely not Dion. Would Rae or Kennedy? I doubt it but I don't know their positions as well as I know Dion's.

john said...

UWH: I saw that page, but I'm not sure how much I rely on that. Like I said, I'm not sure how much I rely on any of these numbers.

I was more interested in seeing how strong Dion's numbers are, and whether he's got a real chance - not something the simple RvI numbers tell us.

Kennedy will endores Ignatieff on a cold day in hell. I don't think Rae is much more likely, frankly.

Anonymous said...

This poll leaves out ex-officio delegates amongst whom Ignatieff has a HUGE lead. This will undoubtly effect the result.

For instance, in EKOS' head-to-head match up between Rae and Ignatieff the two are virtually tied without ex-officios delegates. Once ex-officios ARE included one assumes this will tilt the balance decisively toward the latter.

The Rae/Iggy case also proves that the "second choice" numbers can't be relied on for making extrapolations. Assume that as candidates drop off the ballot, their delegates split according to the same ratio as they chose their second choices (e.g. since 3 times as many Dion delegates chose Rae as their second choice as chose Ignatieff as their second choice, we assume three quarters of all Dion delegates would vote for Rae over Ignatieff on a head-to-head ballot). If we do this for each of the other candidates, Rae should have a 7 or 8 point lead over Ignatieff on the last ballot amongst elected delegates. But he doesn't.

The only way to tell how one candidate will do against another is to poll head-to-head match ups, or ask people to rank each of the candidates and compute how each delegate would vote in each round or with each pairing on a final ballot.

I wish EKOS had done this, but -- alas -- they did not, so we'll all just have to wait for the first week of December.

CuriosityCat said...

This poll plus other public polls of voters generally and Liberals in particular, plus Ignatieff's seemingly weekly series of shooting-himself-in-the-foot plus some sober reflection by all concerned, will see a sea change in how the ex officio voters really vote on the first ballot.

Ignatieff is shedding support like a moulting canary in a heat wave.

How many of the ex officio delegates (who are not legally committed to vote for any candidate) he is shedding on a daily basis, must be a reall concern for the Ignatieff camp right now.

These polls and his missteps really lead to one conclusion: The first ballot is no longer a real test of strength, but simply a formality to be gotten over with as fast as possible. The second ballot is now the king-maker ballot.

And Ignatieff's shot at royalty on that second ballot is diminishing rapidly ...