Austin Contrarian extended the analysis of the chart of transit usage and density I posted before I left. He noticed the same weaknesses in the data that several commenters did—the absence of some cities, and the arbitrariness of using city boundaries which are small in some cities (like DC) and large in others (like Houston).
AC solved the problem by computing his own metrics. Better than standard density (under which LA is denser than NYC, enabling anti-urbanists like Randal O’Toole distort facts to make points), he used the better weighted density. That correlates very strongly with transit ridership.
The best correlation of all, though, was in the ratio between standard density and weighted density—the “the degree of clumpiness” of the population.
In all three graphs, one thing stands out: Washington, DC is still an outlier with higher transit ridership than other cities of similar density (standard, weighted, or ratio).
Thanks to Dan Miller for pointing this out while I was away.