Friday, February 25, 2011

Predicting the race, what if Kevin Falcon is really disliked by party members?

Working through the information from different campaigns, I still have a fair degree of confidence that based on the data I have access to my current prediction is correct.    That said, I have enough information that contradicts some of my previous assumptions that I have worked up another scenario.

In this scenario Kevin Falcon is unpopular with the existing party members and his supportive MLAs are unable to bring a strong win for him in their ridings.   I am moving a lot of that support to Christy Clark.  

I am leaving the sign ups roughly the same but I am assuming that Clark is better distributed than Falcon.  

So working through with this I end up the following:

  • Clark  3200 (3050 - 3350) - 37.6%
  • Abbott 2350 (2250 - 2450) - 27.6%
  • Falcon 2100 (1950 - 2250) - 24.7%
  • de Jong - 850 (800 - 900) - 10.0%

This scenario leaves Clark at 1051 votes out of the winning spot - even if she were to get all of de Jong's support she can not win on the second ballot.  The shifts I made on the spreadsheet give George Abbott by 250 points without changing anything for him.   His support is very responsive to any changes to assumptions about the existing membership and people not signed up be another campaign.

I can not make the math work to make Falcon unpopular enough to give Clark more than 40% on the first ballot.   He has 19 MLAs, 4 Senators, 1 former MLA and 17,500 sign ups.  He has a long list of business leaders as supporters.   This has to count for some weight in any calculation.  

The only way I can get Clark to more than 40% is  is if most the bogus members were from Falcon and he concentrated his vote in five ridings.   I have no data to back that idea, no one has given me anything to say this is case.

Second count - I am assuming the de Jong support splits 400 Abbott, 250 Clark and 200 Falcon, it actually has little impact how I shift them within the realm of realistic on the final result:

  • Clark - 3450
  • Abbott - 2800
  • Falcon - 2300

We need to go to a third count with Falcon's support going to the others.   Here the math get very interesting - how many Falcon supporters only had two preferences and chose de Jong as their second?   I am going to assume 10% are going to do that.   That move effectively changes Clark and Abbotts' totals to 3575 and  2900 - the 230 dropped votes are defacto split between the two before the count.

There are now 2070 Falcon points that have a second preference.   For Clark to win she needs to score more than a third of the Falcon supporters.   I have no decent estimate on the split Falcon voters will do, but I see Abbott leading Clark but I doubt by a large enough margin.   I am going to assume 60%  this leaves us with a third count that is as follows:

  • Clark - 4800
  • Abbott - 3700

If there were no was no drop off, the third count would look like this:

  • Clark - 4350
  • Abbott - 4150

In my opinion that result is within the margin of error of my assumptions to make it a dead heat.

Each Falcon supporter that has no preference marked for Clark or Abbott helps Clark get closer to a win.  If the second result is as above, if 1500 Falcon points do not have a Clark/Abbott choice, Christy Clark wins even if not single one of the rest of the Falcon supporters vote for her.

If the assumptions in this scenario are correct, I see the result as 90% Clark wins and 10% Abbott wins.

No comments: