You've probably all seen the images of the passenger being forcibly removed from a United Airlines flight, to make room for airline staff flying on standby. The incident caused quite a bit of negative publicity, and quickly wiped out a quarter billion $ of the airline's market value. But did you know that other passengers have also been involuntarily bumped off of flights before - by both United and other airlines? Evidently the other airlines didn't get as much publicity, but it happens more often than you might have guessed. Let's have a look at the actual data ...
But before we get into the data analysis, here's a photo to get you into the 'airline' mood. This is a picture of my friend Mary and her son, flying to Disney in Orlando. Although he looks sweet in this picture, Mary assures me that he has the potential to pitch quite a tantrum, if they tried to forcibly bump him from his flight! ;-)
And now, let's analyze some data! I did a bit of web searching, and found an article on our local news site that had just about what I was looking for. They had a graph showing the rate of people who had been involuntarily bumped off of flights by each airline. They had a fairly good bar chart, but it still left me feeling a little unfulfilled:
I like that they used a horizontal bar chart, since that works well with the long text names. And I like that they sorted the bars, with the longest ones at the top. But my main complaint is that the article's focus was on United Airlines, but the graph was more generic (I actually had to look for several seconds to find United in the graph). So, of course I decided to create my own version, to try to improve it.
I did a bit of searching, and found the actual airline data in a table on p. 34 of a report on the transportation.gov website. I copy-n-pasted it into the datalines section of a data step, and wrote a bit of code to read in each line and parse it (I did cheat a little by manually adding some ':' characters to make the text easier to parse). And here is the chart I came up with - click the image to see the interactive version, with mouse-over text for each bar:
Here are some of the changes/improvements in my graph:
- The bar for United Airlines is red, so it is easy to find.
- The bars for the other airlines are gray, so they don't stand out as much.
- I annotated the rate numbers on each bar, so you don't have to guess what the values are.
- I don't use the informal word 'bump' in my graph (since they don't use that word in the data).
- I added much more info in my html mouse-over text than the original graph had.
Since the data table had 2015 data available, I also created a 2015 graph, so you can compare that to 2016. Looks like United improved a little from 2015, and Expressjet had the worst bump rate both years.
So, purely looking at the data, it appears that United Airlines isn't really so bad when it comes to the rate at which they involuntarily bump passengers off flights ... but they certainly got a lot of attention with the way they physically did it this time. I predict that these rates will be much lower for all the airlines, in the 2017 graph!
So, how do you think airlines should handle things like this? Should they be allowed to overbook flights, and then bump people off (either voluntarily or involuntarily)? When they overbook a flight assuming there will be some no-shows, do they (in essence) get paid twice for the those seats? If they don't overbook, then will all our seats cost a little more? If they overbook and don't have enough seats, how much should they pay in compensation to buy some seats back? (should there be a $ limit at which they can start involuntarily removing passengers?)
Have you ever voluntarily given up your seat, and how much $ compensation did you receive?
Kind of surprised that no one has pointed out that the worst two airlines - expressjet and skywest both fly EXTENSIVELY as 'United Express'. Meaning that United Airlines is in charge of those airlines' ticket inventory and would be accountable for oversale situations. Just like the now famous Dr. dok incident in Chicago was on a 'United Express' flight (republic airlines) - the mainline carrier ultimately bears responsibility in the public eye.
The above date should be correlated with the mainline airline operation they are associated with to get a more accurate overview of performance.
The average traveler booking a 'united express' flight from Newark to Buffalo doesn't understand that they are not actually flying on United Airlines, but rather expressjet airlines DBA United Express.
Bottom line - combine the skywest/expressjet/united numbers and you'll get a more realistic denied boarding figure when looking at United Airlines operations as a whole.
Interesting information! -- as they say "the devil is in the details"
Not that I want to be on the Airline's side, but I think it's also important to show how many "no-show" customers they have per year. If the percentage is significant then it's no wonder they have to overbook, in order for them to have a full plane. And if you change the booking method so that people have to cancel (let's say 10 days before the flight), you ruin the flexibility for the business class travelers, who pay much more for the tickets. Hence ticket prices would have to go up. (At least that would be the airline's excuse to raise prices.)
But the airlines still get paid for the "no-shows", don't they? I would think that as long as they get paid for the seat, they don't care if there's nobody in it(?)
Please work your magic with the horrible graphs in this WP article, particularly the last one
Looks like the Washington Post maybe needs to hire a 'graph guy'!
Although the United incident highlights the practice of overbooking, this time it was not denied boarding or overbooking that failed. It was the attempt to force the passenger to pay for crew scheduling errors and then violent off-loading. Overbooking has some value. Bumping is expected. Violent off-loading is not. Thanks for pulling the overbooking data though. Clarifies that aspect.
I think you need to adjust your format for "Enplaned passengers" -- maybe up to COMMA15 or so. At first I thought the American Airlines data tip was showing 130,000 enplaned passengers, and that didn't seem high enough or correct for the ratio displayed. Now I see it's really > 130,000,000 passengers!
Nice catch eagle-eye Chris! It appears when I read in the values, I used a commaxx. informat that was too short, and it truncated the values. I hadn't scrutinized the mouse-over text as well as I should have, and didn't notice it. I have fixed it now (if you still don't see the correct values, you might need to Ctrl+Refresh your browser, or manually clear the cache to replace the old html file with the new one).
I understand from the reports that the passengers were offered compensation for volunteering to be bumped in the form of travel vouchers. Of course if you're not a regular traveller on a route served by that airline the vouchers are worthless to you. I'm guessing also the companies know that a lot of those vouchers won't be used so maybe if they had to give compensation in the form of cold hard cash it might deter them from overbooking in the first place...
When I was young & naive, fresh out of college, I accepted a travel voucher for my seat. I never had an occasion to use the voucher before it expired. :\
I think overbooking is double dipping and should not be allowed. The airline sold a seat and it belongs to the buyer/passenger at least for duration of the flight. If the original passenger does not show up, they may sell that seat to a stand-by passenger, but then they should be passing the proceeds of that sale on to the original passenger (well, maybe minus some commission for rebooking).
Ahh! - I like that idea.
It seems much more equitable!
Imagine if overbooking was the normal situation for concerts: 'I'm sorry, although you have paid to see this performance we have allowed more people than there are seats to pay us for this entertainment, therefore you will not be seeing your favorite band tonight...'
Doesn't sound quite right in this scenario, so why should it be acceptable for transport providers?