An AV count stops at the point where the leading candidate has more votes (both first-preference and transferred) than the other candidate or candidates left in the race. That much is logical and straightforward: the leading candidate can’t be caught up so has won.

This means, however, that some results will be declared when there are only two candidates left in the count, and some will be declared when there are three or more candidates left. This inconsistency makes it impossible to analyse the results and assess the number of winning candidates enjoying the genuine support of a majority of voters in their constituencies. And yet, this is the whole point of AV: that the winners are supposed to enjoy the backing of a majority.

For example, there could be many seats in southern England where a winner is declared while there are three candidates left in the election, especially if Labour picks up the votes of disaffected former Lib Dem supporters. Let’s say that, after the candidates from the minor parties have been eliminated, the tally is:

Conservative: 48%

Lib Dem: 24%

Labour: 23%

The missing 5% are voters who haven’t indicated a preference for any of the remaining candidates.

At this point, the Conservative is declared the winner, and the count stops. The result is recorded in two ways: 1) as a ‘majority’ for the Conservative: 50.5% (i.e. 48% divided by 95%); and 2) as a plurality for the Conservative (48% of all votes cast).

By contrast, in another seat, the standings of the same three parties could be:

Conservative: 48%

Labour: 25%

Lib Dem: 24%

Here, only 3% of votes have failed to transfer to the remaining candidates, and the Tory can be caught. So the preferences of Lib Dem voters are distributed between the remaining candidates, and the final tally is:

Conservative: 53%

Labour: 41%

Total non-transferred: 6%

Here, the final result can equally be recorded in two different ways: 1) a large majority for the Conservative: 56.4% (i.e. 53% divided by 94%); or 2) a majority for the Conservative of 53% (out of all votes cast).

So in the first seat, the Tory winner appears to have won on a plurality, and in the second, (s)he has secured a majority. However, if the preference votes of the third-placed Labour candidate in the first example were distributed to the top-two candidates, the Conservative would probably be shown to have a majority there, too, as only 8.7% of Labour voters would need to put the Tory candidate down as a higher preference than the Lib Dem to give the Tory 50% of all votes cast: not inconceivable if the Conservative is perceived as a good constituency MP.

There will, however, be many other seats that come down to just two candidates where the winner won’t obtain a majority of all votes cast; e.g. Labour-Tory marginals where not enough Lib Dem voters indicate a preference for either of the two other parties. Therefore, when analysing the results, it will not be possible to distinguish between constituencies where the AV winner genuinely enjoyed the support of only a plurality of voters, and other seats where the winner only appeared to be supported by a plurality but would have obtained a majority if the preference votes of third- and lesser-placed candidates had been transferred as they were in the other type of plurality result.

Consequently, it will not be possible to consistently assess the degree to which AV-generated majorities correspond to real majorities of all votes cast or merely pluralities, because the results will include a variable set of preference votes on a constituency by constituency basis. In fact, the counting method employed will exaggerate the number of plurality results, making AV appear even less successful at generating majorities. This will not inspire confidence in AV and will increase voters’ suspicions about its inconsistencies and non-counted preference votes.

The mainstream parties are likely to exploit this inability to objectively assess the proportion of seats where AV has produced ‘genuine’ majority winners, and will claim that the only result that matters is the AV-defined majority in each constituency: the majority of the votes remaining in play at the final stage of the count.

And this is what is ultimately so insidious about AV and how it will be manipulated to legitimise disproportional election results. If a party wins an outright parliamentary majority under AV – which it will be possible to do on an even lower share of first preferences than the share of the vote under FPTP – it will be able to claim that this equates to a ‘majority of majorities’, i.e. that the party is supported by a majority of voters in a majority of seats. This is of course likely to be factually untrue if one is talking of real majorities: majorities of all votes cast. But if there is methodological uncertainty about what the manufactured AV majorities really correspond to – i.e. whether they reflect real majorities of all votes cast, merely latent majorities of all votes cast, or no majority at all – the parties will conveniently dismiss such quibbles and smugly ascribe themselves a false majority mandate anyway.

All of this means that parliamentary majorities produced by AV will be even more unreliable, contestable and illegitimate than FPTP ones. At constituency level, some AV-defined majorities will not be genuine, while in other seats pluralities of all votes counted will in fact correspond to latent but unrecorded majorities – but the parties will claim they’re all majorities regardless. And AV will have failed in its primary purpose: to ensure that MPs enjoy the transparent support and clear assent of a majority of their constituents.