But collisions overall are down, and severe crashes down 80%:
* https://tirf.ca/blog/fact-vs-fiction-roundabouts/
> They work where each direction has an approximately equal amount of traffic. However, usually, one direction (north/south, say) has a larger share of the traffic.
While it would slightly defeat the purpose in some ways, roundabouts with traffic lights are a thing:
* https://www.traindrivinginstructor.co.uk/roundabouts-with-tr...
So if one way has heavier traffic (at times), things could be signalized to allow (e.g.) that one direction primarily some of the time and the other three directions the rest.
> Finally, they're also (mostly) worse for cyclists and pedestrians.
Some data seems to conflict with this:
> A 1993 study in the Netherlands examined collision experience at 181 intersections converted to roundabouts. Pedestrian collisions (all injury severities) dropped 73% and pedestrian casualties dropped 89%.
> Evaluations in Sweden concluded that single-lane roundabouts are very safe for pedestrians, at about a 78% reduction in injuries, and that multi-lane roundabouts are about as safe as other intersections.
> A 2002 study of collision experience at the roundabouts in Park City, Vail, West Vail and Avon, Colorado, showed 2 pedestrian crashes prior to the roundabouts operating with over 164 million vehicle movements, compared to 1 pedestrian crash with roundabouts experiencing over 282 million vehicle movements.
* https://www.tac-atc.ca/wp-content/uploads/weber.pdf
> With the installation of a roundabout, this study finds that the before-after analysis results in a 40% decrease in all severity injury crashes for all roadway users, a 70% decrease in fatal and serious injury crashes for bikes and pedestrians, and a 15% decrease in total pedestrian bike and pedestrian crashes.