One tip for aspiring data hounds — always ask for weighted data.
What do I mean?
What’s the on-time performance for commuter rail?
In NY, it’s 96%! Huzzah, right?
Wrong, that’s on the basis of trains, not weighted by passenger. Anecdotal evidence suggests rush-hour trains (which are packed with people) have a much higher rate of delay.
By not weighting on-time data by passenger, NY’s on-time performance fails to tell the whole story and fails the credibility test with the most important user group of all — the customers.