Updated 19 November 2013 to use Roger Scully’s data on second preferences
Roger Scully has been writing an interesting series of blogs on how we might improve the electoral system in Wales.
In his most recent article, he looks at the potential for using the Single Transferable Vote system.
I have long supported the STV system as the most effective means of securing proportional representation. It delivers a fair result without the “two tier” system of constituency members plus “top up” members. The main argument against STV is that large multi-member constituencies would reduce the link between voters and their elected member. The danger is that we would end up with lots of “regional members” whereas people feel a closer connection with their constituency members.
My preferred solution for an improved electoral system for Wales would be two-seat STV, which would ensure the Assembly was representative while maintaining a close link between members and their constituents. It would make sense to require all parties to field one man and a woman in each seat where they were standing two candidates, thereby ensuring a much better gender balance in the Assembly (and removing the need for all-women shortlists, the only mechanism any party has used to date which has had any impact on improving gender balance).
The results under such a system would be fairly similar to Roger’s projection for four-member STV:
|Two seat STV||Four Seat STV|
|Labour||42 – 43||40|
|Plaid Cymru||8 – 9||15|
There are some seats where third and fourth preferences might make a difference and of course there could be more minor parties contesting these elections, whereas most very small parties chose instead to seek regional seats. Roger has provided data for Green and UKIP transfers – perhaps unsurprisingly UKIP favour the Conservatives while Greens favour Labour. Labour voters meanwhile favour Plaid Cymru. There is no data on Socialist Labour or BNP: they secured just 2.4% each across Wales, but there may be enough of them to affect the outcome in some seats. Rhondda is the closest result: Labour win the first seat comfortably, but based on constituency votes they would win the second seat by just 27 votes. In a result that close, the third preferences of Conservative voters could determine the outcome. On regional votes, Plaid are ahead by 566 but that is without transferring the votes from Socialist Labour (746) and BNP (476) and other minor parties (483) while again third preferences may also have an impact.
I agree with Roger that STV ties the outcome to popular wishes – not just in terms of a voter’s first preference but also their subsequent preferences. Candidates and parties picking up second preferences will do better than those candidates and parties inspiring people to vote anybody but them. This system also has the advantage of eliminating safe seats: although there are twelve constituencies where Labour would have taken both seats, in 2007 there were no seats where this was guaranteed and only three where it was a possibility of which only one was likely. In 2007 the Conservatives and Plaid Cymru had a chance of picking up a second seat in Monmouth and Dwyfor-Meirionydd respectively.
Compared to four-seat STV, it is clear that this system favours the larger parties (note that the totals for Labour and Conservative are marginally higher whereas Plaid Cymru are lower). But personally I think this is better than the current system where the vast majority of seats are distributed on first past the post and the regional seats are too large to have a close connection with the electorate.
It is worth looking at the results in 2007 as well – an election where Labour was still largest party but fared comparatively badly. Under this system I project that Labour would have secured 35 of the 80 seats. This is a similar result to the existing system (Labour secured 26 out of 60 seats). It would be wrong to use the second preference data from 2011 – for example some of the Labour voters who were sympathetic to Plaid Cymru in 2011 voted for Plaid Cymru in 2007 so you cannot assume the proportions of second preferences would have been the same.
In addition to the difficulties in predicting what happens to transfers, of course this system would have an impact on where parties campaign – with no “safe” seats and a need to secure second preferences as well as first preferences. Their success in campaigning would determine the actual outcome.