Thanks for that link. I wasn't really interested comparing the overall averages, I was interested in whether my anecdotal experiences of Frontier being delayed more than United (for example) on days with weather issues at LGA was typical or unusual. Direct apples to apples as it were. Although I didn't do a comprehensive examination of both departures and arrivals, and the data provided is not exactly detailed, I think what I did find with a first pass is kind of interesting.

I pulled daily Departure and Delay data for the route I was looking at, DEN->LGA in the approximately 10-11AM time (United was more variable in their scheduled times) for F9 and UA for every day in 2012 for which data was available (335 days, through the end of November). I found that neither carrier used the "extreme weather" code as a cause for delay, so I looked at the NAS (non extreme weather) code. But that sometimes was coded with a longer delay than their overall reported delay, not sure what that means. So I took their overall reported delay time (actual minus scheduled departure times) and subtracted out the other delay categories (carrier-specific issues, late aircraft) to get an approximate "delay related to weather when the plane is actually there" value. (Of course this doesn't include things like sitting on the tarmac.)

To get a sense of the more extreme cases I decided first to look only at delays of 30 minutes or more for each airline. Here's what I found.

In the first 11 months of 2012, United reported 8 such days. Frontier had delays of 30 min or more on 5/8 of those days (63%). Based on a paired t-test, the average delays on each airline on those 8 days were not significantly different: UA Mean=49 min and F9 Mean=37 min (Mean Difference = 12 minutes, p=.29).

Over the same 11 months Frontier reported 31 such delay days, 5 of which (16%) were shared by United. Average delay times were significantly different: UA Mean=13 min and F9 Mean=62 min (Mean Difference = 49 minutes, p<.001).

Based on a chi-square test, the distributions of shared days (5/8 vs. 5/31) were significantly different at p=.001.

My conclusions are: 1) that on comparable DEN->LGA late morning flights, Frontier reported nearly 4x more (31 vs. 8) delays related to weather (not carrier-specific or due to late-arriving planes) of 30 minutes or more than did United; and 2) that the average difference between delay times for the two airlines was 4x longer (49 vs. 12 min) for Frontier on its delay days than for United on its delay days.

As I said, the data are a little messy and the categories broad, and maybe departure time isn't the best metric and etc. etc., *but* overall it looks to me like these data do indeed support my impression. When United had to delay departure at least half an hour due to weather, which happened less than once a month, nearly 4 out of 6 times Frontier had to also. But when Frontier had to delay departure at least half an hour due to weather, which it did roughly 3x/month, United had to do likewise only about 1 out of 6 times.

What remains unclear is the reason for the difference.