The most popular topic is not necessarily the most important topic

The most common approach for prioritising topics in retrospectives is some form of popular vote with an underlying, implicit assumption that what is most popular is what should be discussed.

Is this assumption trustworthy?

• Would I not vote for something because I find the topic uninteresting?
• Would I not vote for something because I find the topic uncomfortable?
• What I be more likely to vote for something because it had happened to me recently?

The question should not be "What do you want to discuss?" but rather "What are the most important things to discuss given our agreed target outcomes?"

This means being clear on what those target outcomes are.  See Goal Driven Retrospectives.

This also means that we need collect data before the retrospectives, not just during.

I've started to suggest that teams have check sheets to record issues with a running tally each time the issue occurs.  By the time the retrospective occurs, you can look at what has occurred the most often versus what has just happened to have occurred right before the retrospective.