Okay, I'm done with this conversation since no one is making new arguments. People who believe that AA provides equally good customer service whether or not they are flexible in situations like this will not be persuaded otherwise, even in the face of examples of both AA agents and other airlines who manage to be reasonably accommodating without delaying hundreds of other passengers willy nilly. If people want to maintain this belief, I can't stop them, but I think it's symptomatic of the general Stockholm Syndrome that members of this board consistently exhibit, defending even the worst behavior of their chosen airline.

Having said that, I'm of the general opinion that some data trumps no data almost all of the time. Virgin's on-time performance is sufficiently better than AA's that you'd have a hard time convincing me that being really anal about gate arrival cutoff has any meaningful effect on on-time performance. You can feel free to dismiss the data as "garbage" since it doesn't support your position, but it would be stronger to have some data that supports the counter position rather than just trying to bring us back to the realm of rank speculation. It's quite easy to take any particular data point and explain why it doesn't apply perfectly in a situation; it's quite another to present evidence that more persuasively argues in the other direction.

Like I said before, the interesting thing here would be if the DOT started tracking % of passengers who were delayed as opposed to % of flights. This would give airlines a stronger incentive to balance the costs to people missing connections in these situations with those already on the plane. But as it is, the mission is clearly to get the flight out on time regardless of knock-on effect on passengers (or customer satisfaction).