Originally Posted by
ijgordon
6-12 months ago I did a similar exercise on several routes where there were good "apples to apples" comparison flights on other airlines, specifically noting to compare flights departing at around the same time, because there can be big variations on a flight leaving, say, JFK at 9am vs. 6pm. Domestically, I don't think you need to control for aircraft type as I believe cruise speeds are similar on narrowbodies and the distances short enough to not matter much (but if you're comparing, say, a 767 to a 777, the latter is noticeably faster and it matters on a long-haul).
My takeaway (and I don't have the details) was that DL padded by about 5 minutes or so on average (but not every flight).
I think if you had full access to OAG you could do a real comparative analysis of flight schedules across the network, but otherwise it's just anecdotal.
A couple times when there were schedule changes involving changes of aircraft type but not departure times, I noticed that the scheduled flight times (and arrival times) were also changed by a few minutes.