This post originally appeared on Politics in Spires.
On Monday David Cameron and Alex Salmond signed an agreement allowing the Scottish government to hold a referendum on independence. The Scotland independence referendum will be the latest in an ever growing list of referendums held in the UK that began in earnest with the Blair government. Even when referendums have not been held there have been calls for them on a variety of topics: from reforming the House of Lords, to Britain’s membership of the EU. With the electorate increasingly consulted directly on a broad range of issues you would be forgiven for thinking that the use of the referendum had brought about a golden age of democracy. You would however, be wrong.
The Scottish independence referendum will be on a single yes or no question to whether Scotland should become independent, leaving out a proposed second ‘Devolution Max’ question, which would grant the Scottish government considerable further powers whilst remaining part of the United Kingdom. The exclusion of the Devo-Max question has disappointed many because, as Lesley Riddoch points out in the Guardian, public opinion polling suggests that this is actually the option that Scots want.
That there will only be a single question is a good example of something well known to those that study social choice theory: the power that someone with control over the agenda has over the outcome of voting. The power of the agenda setter has been long known: In his book Theory of Voting Robin Farquharson relates a tale of Pliny the Younger attempting to manipulate the Roman Senate by changing the usual voting procedure. The full extent of the possible effects of agenda control was first noted by a colourful 18th Century French mathematician, the Marquis de Condorcet who in his famous paradox noted that in a situation with a tie between three options a voter can determine the outcome of an election by voting insincerely on one of their preferences. In 1976 Richard McKelvey went much further, in a complex series of mathematical proofs, establishing that given control over a voting agenda it was possible to bring about almost any outcome the agenda setter desired.
That there will only be a single question on the Scottish independence referendum is a good example of the power of agenda control in action. Unless public opinion changed drastically by the time of the referendum the Devo-Max option would have been a clear winner if it had been included. Excluding it suits the purposes of both those who favour complete independence, who will hope they can convince Devo-Max supporters that full independence is better than the status quo and those who oppose a further transfer of power away from Westminster. It should not be surprising then that David Cameron and Alex Salmond, who completely disagree on what the outcome of the referendum should be, found common cause over a single question fairly easily.
The power of agenda setters to manipulate the outcome of referendums helps to explain a frequently observed paradox of referendum voting: high levels of public support for general change frequently don’t translate into support for specific change. It helps explains why in the UK, where support for a more proportional electoral system has hovered above 60% for 30 years, 68% voted in favour of keeping the First Past the Post system. Or why in 1999 Australia, where support for becoming a Republic has also long been around 60%, 55% voted in favour of keeping the monarchy. In both cases the referendum question was largely determined by a Prime Minister who was strongly in favour of the status quo.
The power of agenda control means we should be wary of those that call for the use of referendums. As we saw in the debate over Lords reform, the strongest voices in favour of a referendum are often those that oppose the very change they offer. These are not benevolent democrats, but Machiavellian calculators. Those that decide to hold referendums will claim that they do so in order to provide democratic legitimacy to an issue. My own research, which I’ve discussed on this blog previously, shows that strategic considerations by governments, designed to maximise electoral success, are a better predictor of when referendums will be held.
Referendums look a lot like democracy. It may seem strange to say that a process where voters are consulted about their views on an issue is not in fact democratic. Unless the issue at stake is truly binary however, the outcome of a referendum tends to reflect not the views of the electorate but those of the person who set the question. Although no electoral system is perfect, and the UK’s First Past the Post method is worse than most, parliamentary elections are a better expression of democracy because unlike a referendum, there is no clear agenda setter: the agenda is contested between parties and candidates, the media, and, increasingly, via the internet, the voters themselves. Those who think of themselves as democrats would do well to remember these issues. The referendum is not an instrument of democracy. It is a tool of politics.
Chris Prosser is a DPhil student in Politics at St. Catherine’s College, Oxford.