The NYT has a really good piece on why referenda are problematic instruments: Why Referendums Aren’t as Democratic as They Seem
Though such votes are portrayed as popular governance in its purest form, studies have found that they often subvert democracy rather than serve it. They tend to be volatile, turning not just on the merits of the decision but also on unrelated political swings or even, as may have happened in Colombia, on the weather.
Voters must make their decisions with relatively little information, forcing them to rely on political messaging — which puts power in the hands of political elites rather than those of voters.
Voters face a problem in any referendum: They need to distill difficult policy choices down to a simple yes or no, and predict the outcome of decisions so complex that even experts might spend years struggling to understand them.
Voters typically solve this problem by finding what the political scientists Arthur Lupia and Mathew D. McCubbins have termed “short cuts.” The voters follow the guidance of trusted authority figures or fit the choice within a familiar narrative.
When a referendum is put forward by the government, people often vote in support if they like the leadership and vote in opposition if they dislike it, according to research by Lawrence LeDuc, a political scientist and professor emeritus at the University of Toronto.
“A vote that is supposed to be about an important public issue ends up instead being about the popularity or unpopularity of a particular party or leader, the record of the government, or some set of issues or events that are not related to the subject of the referendum,” Professor LeDuc wrote in a 2015 paper.
In Colombia, for example, most regions that voted for President Juan Manuel Santos in 2014 also voted for the peace deal, and vice versa.
All of which I think is on target. I would recommend the whole thing.
I also now have a new phrase to describe referenda: ”This isn’t democracy; it is Russian roulette for republics” (Kenneth Rogoff).