One of the myths that is widely peddled about AV is that it ensures that seats are won with the support of over 50% of the electorate. It doesn’t, as I hope I’ve managed to convey in numerous interventions in comments on this site.

Another myth is that it eliminates the need for tactical voting. It doesn’t. For a start, the last preference voters indicate on their ballot paper under AV will almost certainly be effectively a tactical vote: out of those candidates that have a chance of winning, this is the one you’d most like to do so. In terms of the actual effectiveness of your vote, as opposed to any emotional satisfaction you might gain from being able to vote with your heart, you might as well not bother indicating smaller parties as your higher preference, if that is what you are inclined to do: you might as well just go straight to the one acceptable party you think could win, just as you would under FPTP.

The fact that AV, just like FPTP, has the effect of denying smaller parties a chance of winning any significant representation in itself creates a dynamic towards tactical voting: you know your preferred party can’t win, so you end up voting tactically all the same.

Equally, the somewhat counter-intuitive mechanism of AV actually creates openings for tactical-voting strategies that seem even more unfair and absurd than the more obvious ones under FPTP. I think everyone realises that the Lib Dems in theory stand to gain most from the more obvious type of tactical voting under AV: Labour and Tory voters putting the Lib Dems down as their second choice, so that if their party is eliminated and the Lib Dems are still in the race, the Lib Dems stand to gain – typically, this scenario might benefit the Lib Dems in southern England or Scotland.

However, there is a more perverse scenario for pro-Lib Dem tactical voting in areas where the Lib Dems would typically finish a mediocre third under FPTP, such as in the North of England and the Midlands. What this involves would be Labour voters putting the Lib Dems down as their first preference – even though Labour is more popular in those areas than the Lib Dems – in order to get the Lib Dems into the final pairing of two parties to which votes for the other parties are transferred in the AV system. This would ensure that the second or later preferences of those who had continued to put Labour down as their first choice would almost exclusively be transferred to the Lib Dems, ensuring that they, rather than the Tories, would win.

In these cases, the opposite strategy – Labour voters putting Labour down as their first choice – carries the risk of generating a Conservative victory, because there might be enough Lib Dem second preferences that would go to the Conservative candidate rather than Labour, especially if the Lib Dems and the Tories were still in a coalition.

Take the West Midlands ‘region’, for example, where the Tories gained 39.5% of the vote on average in the 2010 election, compared with 30.6% for Labour and 20.5% for the Lib Dems. Assuming the popularity of the parties remained roughly the same in 2015, and that the votes for more minor parties were transferred evenly between the three larger parties via the AV mechanism, then all it would take for a Tory victory to be secured would be for around one-third of Lib Dem voters to switch their allegiances to the Tories in their second vote. By contrast, if just under one-sixth of Labour voters (i.e. around 5% of the total vote) transferred their first preference to the Lib Dems, then it would be the Lib Dems, not Labour, that would go into the final pair-off with the Tories. And assuming that virtually all Labour votes first-preference votes were transferred to the Lib Dems, the Lib Dems would win. This dynamic would be even more powerful in seats where the gaps between the parties were closer than these average figures.

This tactical-voting scenario works even in seats where Labour would otherwise win on the basis of a plurality under FPTP, e.g.  in a ‘region’ such as Yorkshire and the Humber, where Labour polled 34.7% in 2010, compared with 32.5% for the Tories and 23.0% for the Lib Dems. Assuming a slight swing from the Conservatives and the Lib Dems to Labour, you’d still only need just under a quarter of Labour voters to mark the Lib Dems as their first preference to get the Lib Dems into the final pair and ensure a Lib Dem victory – rather than run the risk that Lib Dem second or subsequent preferences would hand a win to the Tories.

These counter-intuitive tactical-voting scenarios derive from the fact that not all the second preferences of voters are counted under AV, only those from voters whose higher-preference parties have been eliminated. Hence, if the Lib Dems finish third, it will often be the second preferences of Lib Dems that will be decisive, and this could carry the day for the Tories. So by voting Lib Dem as their first choice, Labour voters could try to pre-empt a Tory win. And this is a low-risk strategy, because if it fails, then Labour are still in the race, and more second preferences of people who’ve voted for the Lib Dems as their first choice will be transferred to Labour, as these will include many actual Labour voters.

This is just one example of how unsatisfactory AV is. Tactical voting will continue, but it will produce even more perverse effects than FPTP.

Read more about the AV referendum in OurKingdom's Referendum Plus section.