AV for Referendums? Under no circumstances!!


AV (the Alternative Vote) should NEVER be used to decide among policy options offered in a referendum. Why not? Because the policy option that has the greatest overall support could well be rejected in an AV count.

Consider this simple example, where there are three options (A, B and C) and 100 voters. The numbers of voters marking combinations of preferences might be as follows (with irrelevant preferences omitted):
36: A, C
33: B, C
31: C.

None of the three options A, B or C has a majority of the first preference votes, so under the Alternative Vote counting rules, the option with the fewest votes, C, would be eliminated.
But it is perfectly obvious that C is the option that has the greatest level of overall support among the voters.

There are other counting systems that can be applied to such preferential ballots, but they all have their own problems, e.g. resolution of cycles in Condorcet counting. And there would likely be major political problems with some of the possible outcomes, e.g. a winning option which had very little first preference support.

Fortunately all of these problems can be avoided by adopting the “two question” solution. I have used this approach when a group of 146 Proprietors had to make legally binding decisions about the property they owned in common and they were faced with having to decide among three options for the particular aspect of the management of the common property: “No Change”, “Change A”, “Change B”.

These three options resolved into two simple questions, both of which could be given direct “Yes / No” answers.
In the property management case the questions were (in generic terms):
Q1. “Change” or “No Change”.
Q2. If a majority votes for “Change” in response to Question 1: “Change A” or “Change B”.

It was made clear that every voter could (should) vote on Question 2 irrespective of their answer to Question 1. By adopting this approach, clear and unambiguous answers were obtained to show the wishes of the voters.

There was no possibility of problems like those that can be experienced with the Alternative Vote (defeat of “everyone’s second choice”), there were no other technical complications, and there could be no argument about the interpretation of the results.

The options to be considered in the likely Referendum on EU membership could similarly be presented in two separate direct questions and this two-question approach is the voting method that should be adopted for any such Referendum.

James GIlmour

But certainly not FPTP

We welcome comment and discussion and appreciate both James Gilmnour's point and the anonymous response to it, but we can all agree that a FPTP ballot, which would let one faction win with only just over a third of the votes, would be completely inappropriate.