Carl Schmitt and AV Dicey are two of history's most influential constitutional theorists, and they offer two of history's most influential accounts of referendums. In most respects, their approaches to referendums are in direct opposition to each other. On Schmitt's view, the purpose of referendums is to acclaim executive actors. On Dicey's view, the role of referendums is to constrain them. Despite disagreeing about whether referendums should acclaim or constrain the executive, Schmitt and Dicey agree that an agenda-setting role for representatives in referendums is inevitable. This paper argues that, in the UK context, if Schmitt and Dicey are right about the necessary agenda-setting power of representatives in referendums, then the accounts of referendums they each offer must be two sides of the same coin. Given the dominance of the executive over the legislature in the UK and the uncodified nature of the constitution, referendums are processes that necessarily both acclaim and limit the executive.

Recent referendums show that autocratic regimes consult voters even if the outcome is a foregone conclusion. They have been doing so with increasing frequency since Napoleon consulted French citizens in 1800. Why and when do dictatorial regimes hold referendums they are certain they will win? Analysing the 162 referendums held in autocratic and non-free states in the period 1800–2012, the article shows that referendums with a 99% yes-vote tend to occur in autocracies with high ethnic fractionalization and, in part, in sultanistic (tinpot or tyrannical) regimes, but generally not in communist (totalitarian) states. An explanation is proposed for this variation.