Well, I suppose the strength is that it potentially engages the citizens and legitimizes a change in policy, a change in electoral system, or not, as the case may be. The difficulty with referendums is that quite often the thing that people are asked to vote on is not simple. Adopting a new electoral system, for instance, is not simple. Exactly what are the implications? You can find one expert to tell you one thing, and you can probably find another expert to tell you something else, so the voters have to decide between the points of view being put forward by those who say you should vote yes, and those who say you should vote no.
Some voters won't bother to sort out who they want to follow; they'll identify someone they don't like, like the government, and say, “Well, if the government says this, we won't do it”. And that's common in many referendums, and particularly those referendums that are lower in salience, that are not so important to people, and therefore they don't get to know the issue.
The referendum in Scotland on independence had a huge turnout and a very high level of engagement with the topic. I think referendums you've had in Quebec were fairly similar. We've had some referendums like that. We've also had referendums with relatively low engagement. We had two referendums on European treaties when the voters gave the wrong answer and said no, so we had the referendum again the following year with almost no change, and the voters then said yes. Turnout was higher, campaigning was more extensive, and people had time to think about it. If you do have referendums, I think you need an awful lot of resources going in to inform people. You've also had a referendum, I think, in British Columbia on adopting the single transferrable vote, which was carried overwhelmingly, but not overwhelmingly enough, so you've got that experience to look at.