In response to this item two days ago, on the substantial shrinkage in total capacity on U.S. domestic air carriers -- fewer flights, smaller planes, -- Amy Cohn, of the College of Engineering at U Michigan, writes in to explain one of the non-obvious consequences of ever-fuller flights:
There's an interesting corollary to all this: Fuller flights also make delays even worse than they already are. Ironically, cancelling a flight is sometimes the best thing that you can do for system delays - reduce the congestion and everything runs smoother. You used to see that happen a lot. Cancel the 10:00 to DCA and move everyone to the 11 and the 12. Putting a passenger on an on-time 11:00 is often actually better than a delayed 10:00, plus it helps gets the system back on track. But with flights so full, you can't do that - there probably isn't both an 11 and a 12 and even if there are, they don't have empty seats (I watched this happen in STL last night). So the airlines delay instead of cancel, which keeps the congestion levels high, which delays everyone else as well...
Makes me not so optimistic about my flight home...
Another reader, who works in the air travel business, writes with a complaint about this chart from the original item. First the chart, then the (valid) exegesis:
The reader's note:
The third chart (System Load Factor) in "Why Your Plane is Always Full" is, to the casual observer, misleading in that it appears as if load factors for some airlines have nearly doubled over the past ten years. In fact, the real change in load factors has been about 10%--significant, but not as dramatic as a casual inspection of the chart would indicate.
This misleading appearance is, of course, the result of baselining the y axis of the chart at 63% instead of 0%. A 0% baseline would put the increase in better perspective.
I blame this common error on Microsoft Excel. For years, I have taught beginning, intermediate, and advanced Excel to continuing education students at the University of New Mexico. When I introduce the subject of charting I point out to my students that using Excel charting is like giving liquor and car keys to teenagers: the outcome is fraught with peril. Depending on the charting defaults, Excel may automatically make scaling decisions like the one in the subject chart, baselining the scale so that the chart shows only the 'relevant' range.
I regularly show students charts like System Load Factor and ask them to tell me the magnitude of the change. Most will conclude that load factors, or whatever, have doubled until I point out that the y baseline has been jiggered. I then show them how to reset it to zero and to re-evaluate the visual message of the chart. (BTW, my most reliable source for misleading charts and graphs is USA Today where visual appeal always trumps meaningful display of data).
I consider Excel to be an extraordinarily useful program, perhaps the single most important piece of software ever developed for the personal computer. But, its charting features always have been graphically weak and capable of producing misleading results. Some of the charts available in Excel (e.g., donut charts) are more than worthless; they are downright misleading.
By the rules of good charting, whenever a baseline is set to something other than zero, the y axis should show a zig-zag break near the baseline. Excel charts do not do this automatically and, in fact, provide no easy-to-use tools to create such a break.
The choice of scaling and baselines in charts can have a significant impact on the viewer's impression of what has happened. For example, normal weekly fluctuation in the stock market can be made to look like a major increase or decrease just by playing with the scale and baseline.
A personal observation: I'm in the aviation business so this particular post was of more than passing interest. I pay attention to load factors, not only industry-wide, but also on the aircraft in which I am riding. I can't remember the last time I was on an aircraft that was less than 95% full. Somewhere, there must be aircraft that are only 70% full but I never see them.
We want to hear what you think about this article. Submit a letter to the editor or write to firstname.lastname@example.org.