Editors note: Sometimes I can’t help but get academic and nerdy; but stick with me, the results are good. There’s good stuff in the academic community, and we can apply it directly to your travel to make it better. I don’t know of anyone else doing anything like this, so here we are.
Introduction
There’s an interesting statistics thought experiment that comes up in academia called The Monty Hall problem. The gist of the problem is:
- You have three doors with something behind each door, 2 doors have something lame and 1 has something great
- You choose a door but don’t yet know the results
- The game-master tells you that one of the doors you didn’t pick has a lame prize, and shows you which door
Ok, so there are two doors left: The one you picked and the other door. Unless you’re trained in statistics, you probably think you’ve got a 50% chance that your door has the great prize and a 50% chance that the other door has the great prize, because there are only two left. But, the math behind the Monty Hall problem says that your door is 33% likely at that point to have the great prize, and 67% likely to have the lame prize. (See the Wikipedia page for the math behind the result if you’re interested.) In other words, the other choice is now twice as likely to be the best choice, so choose it if you can!
Applying This Result to Flights
We can apply this result to airline delays with some fuzzy mapping: one door is your on-time departure (your original choice, a delayed flight might be un-delayed and is thus still an option), one door is your delayed departure, and the third door is a an alternate flight.
Based on the math behind the Monty Hall problem, if you’re told that your original flight is delayed, then switching to an alternate flight is more likely to get you to your destination without a late arrival; twice as likely all things being equal (which they’re not). If you’ve ever experienced rolling delays on your original flight, you’ve got some intuitive feel that switching to another flight is probably less likely to lead to an arrival delay too.
Making it Real
There’s a problem with that analysis though: It’s highly unlikely that you’ve got an alternative flight to switch to that leaves at the same time as your original flight. So, to make this actionable for real-world scenarios, we’ve got to factor average delay time into our analysis. To do that, I downloaded the last 12 months worth US airline flight on-time data for a deeper-drive.
First, let’s assume that your airline posts a delay of 45 minutes or longer. In the last year, this is what each major carrier’s average arrival delay looked like:
| Operating Airline |
Average Arrival Delay, August 2021-July 2022 (For Departure Delay ≥ 45 Minutes) |
|---|---|
| AA | 2 hours 13 minutes |
| Alaska | 1 hour 36 minutes |
| Delta | 2 hours 1 minute |
| Frontier | 1 hour 51 minutes |
| JetBlue | 2 hours 17 minutes |
| Spirit | 1 hour 49 minutes |
| SkyWest | 2 hours 21 minutes |
| Southwest | 1 hour 23 minutes |
| United | 1 hour 53 minutes |
So when your airline posts a delay of at least 45 minutes, if you’ve got an alternate flight that leaves within an hour and a half or so, you should switch to that alternate flight (especially if your flight is operated by SkyWest).
Next, let’s assume your airline posts a delay of 90 minutes. In the last year, you’re looking at an average arrival delay of:
| Operating Airline |
Average Arrival Delay, August 2021-July 2022 (For Departure Delay ≥ 90 Minutes) |
|---|---|
| AA | 3 hours 23 minutes |
| Alaska | 2 hours 35 minutes |
| Delta | 3 hours 19 minute |
| JetBlue | 2 hours 54 minutes |
| Frontier | 2 hours 54 minutes |
| Spirit | 2 hours 49 minutes |
| SkyWest | 3 hours 36 minutes |
| Southwest | 2 hours 19 minutes |
| United | 2 hours 59 minutes |
The conclusion from this one: If your departure delay is posted as 90 minutes or later, switch to an alternative if you can get one in the next three hours or so.
Finally, let’s look at the data by major airports instead of by airline, sorted by the total number of delayed flights (these major airports are also the airports most likely have alternative flights):
| Airport | Average Arrival Delay, August 2021-July 2022 (For Departure Delay ≥ 90 Minutes) |
|---|---|
| DEN | 1 hour 42 minutes |
| ORD | 1 hour 52 minutes |
| DFW | 1 hour 49 minutes |
| ATL | 1 hour 43 minutes |
| MCO | 1 hour 50 minutes |
| CLT | 1 hour 42 minutes |
| LAS | 1 hour 37 minutes |
| LAX | 1 hour 50 minutes |
| PHX | 1 hour 40 minutes |
The statistics aren’t very different for other major (top 50) US airports. However delays are much more likely to extend beyond two hours at small airports, where you likely don’t have another option anyway.
And for my last analysis, I looked at the reason for the delay when it was available. In cases where the data is available, the longest delays are caused by (from the biggest contributor to the smallest):
- Carrier delays (crew problems, mechanical, etc.)
- Late aircraft delays (delayed inbound flight)
- Airspace delays (ATC traffic management programs, etc.)
- Weather delays
tl;dr
The internet: “Ok poindexter, enough with the nerdy stuff, how about a summary without all the goo?”
MEAB: Sure thing boss, also here’s the data in CSV form in case you want to be a nerd too:
- If your flight posts a delay of 45 minutes or longer, switch to an alternate if there’s one available in the next two hours
- If your flight posts a delay of 90 minutes or longer, switch to an alternate if there’s one available in the next two and a half hours
- If you’re flying out of a major airport, a delay isn’t likely to carry on past two hours
- If you’re flying out of a small airport, that delay is probably going to be a long one, sorry
- If the reason for your delay is a carrier or late inbound aircraft issue, the delay is likely to be longer than for weather or other reasons
Happy Tuesday friends!
When United Express inevitably has a delay for something like this, switch flights (trust me, been there).