Alternate Vote 2015

To test the proportionality of Alternate Vote, it was necessary to estimate how voters might list their second, third, fourth and fifth choices on a ballot. To do this, I started with the results of the 2015 election. Then, I took two studies of voter’s second choice by party, Ekos (2015) and Nanos Research (2015). (Disclaimer: Neither firm has been involved in this analysis).

A similiar methodology was used by Éric Grenier at threehundredeight.com in a simulation published at CBC in 2015.

I averaged the results of the two studies above, and then built an AV election simulation. It works by taking each riding in Canada in turn. If the top rated candidate received over 50% of the vote, then they were declared elected in that riding. If not, the candidate with the least number of votes was dropped off the list, and that candidates votes were re-distributed based on the second choice percentages among the remaining parties. This process was repeated until one candidate received over 50%.

The result showed the democratic deficit is much higher with AV than with FPTP. Remember that the democratic deficit is the number of seats that are occupied by the “wrong” party based on proportional representation, so we want this number to be as low as possible. Under FPTP, the democratic deficit in 2015 was 50.1. Under AV, the simulation suggests that the democratic deficit would be 92.6.

Capture

It is difficult to underscore how bad this score it. Almost one-third of the seats in Parliament would be occupied by a party that didn’t “deserve” this seat based on a proportional split.

One of the consequences of implementing this system is that small parties will be systematically pushed out of contention. Ultimately, this will lead us towards a two party system, as we see currently in the US.

In order to demonstrate how close this is to chaos, I created a simulation which, instead of looking at the 2015 votes, simply went through all of the ridings and picked a winner at random. I repeated this simulation 500 times in order to derive some averages, and the results were quite surprising to me. The average democratic deficit for random winners was 113.7 seats, which is not much worse than AV.  Of course, the governing party was different in all of these simulations, but random selection tends to elect parties which ran candidates in all ridings, so the major parties are well represented in a random selection.

Random2015

This result is astonishing, and I would welcome others to do the same simulation. I think we can all agree that picking MPs randomly has to be the worst possible way to run an election. Yet, the result is not much worse than the simulated results of Alternate Vote, in terms of its relationship to the wishes of the people.