Cohesion, clarity, stability

In addition to  criteria such as transitivity, independence of irrelevant alternatives and neutrality that are desirable for voting rules according to Kenneth Arrow, one could think of other criteria that are especially applicable to sovereignty referendums such as the Brexit one. These criteria could be those of cohesion, clarity and stability.
Cohesion. A yes/no self-determination referendum can be a cause of great division in internally diverse societies. Taking the Quebec referendum of 1995, the Scottish referendum of 2014 and the Brexit of 2016 as examples, all produced a very close result: it seems that these referendums with two options tend to divide the electorate into two halves and produce very crisp and emotional campaigns. The winning option defines a model of society for which almost half of the population has explicitly voted against. In what situation this immense minority remains in terms of risk of discrimination and discomfort, should be a cause for concern.
Stability. Several observers have pointed out the risk of contagion or domino effect, both internal and external. Some might argue that this should not be a problem, since holding more referendums can only be even more democratic. However, it is difficult to find advocates of the secession of a territory that admit the right to become independent of important parts of this territory. The existence of waves of independence and referendum processes suggests that there are imitation effects, which can alert leaders of powerful powers on the international scene, even when some referendum might be desirable to address a serious problem of coexistence or human rights. Partitions and reunifications are not like playing with lego blocks, and have consequences for long term investments made by ordinary individuals and families (personal bonds, labour contracts, social security, homes, professional degrees…).
Clarity. Following the last referendum in Quebec in 1995, the Supreme Court of Canada asserted that it was necessary to take into account the principle that the options presented to the electorate should be clear and avoid confusion, and that the consequences of whatever is decided must be clear and what is decided must also be approved by a clear majority. The Clarity Act subsequently approved by Parliament applied these principles to the specific case of Canada. Note that there is some conflict between posing a clear question and presenting clearly the consequences of what is approved. Making a referendum with two or even three or more discrete options on something that is actually a continuum (the degree of sovereignty) and that does not depend only on the electoral body can induce a sense of "false clarity." There is a risk of "approving" something that is actually pending negotiation.
The solution is not simply to increase a number of seemingly simple options to three or more, because then the question still gives the false impression of simplicity (it would have been difficult to know exactly what devolution max  in Scotland meant without a detailed prior agreement).
By expanding the number of "reasonable criteria" we also increase the number of trade-offs that are typical of social choice. For example, it is difficult to achieve clarity without undermining cohesion. After all, a brief and dichotomous question is very clear, but it facilitates polarization in two opposing blocks, and if we look at the British case, it does not seem to have led to stability.