Test: Are you a good manager?
To read management information correctly, you often need good visual judgment. At least, it seems that way. Are you good enough to interpret the visualizations that you are expected to? Take the test and see for yourself!
The New York Times recently used a chart based on area size to illustrate how the market value for the largest U.S. financial institutions has changed over the past year. The hollow square represents the 2007 value while the filled square stands for the current one. Now, try to guess how much the market value for Citigroup, Bank of America, Berkshire Hathaway and AIG has changed in percent over the past year.
If you enter your estimates in the following fields and click ‘Evaluate’, you will learn how good your visual judgment really is.
I recently took this test along with a few friends and colleagues. On average, we were 24 % off target. That’s too much if you ask me. This test clearly backs the existing arguments against using areas in charts, which were first introduced – alongside time series, bar and pie charts – by William Playfair in the 18th century.
One stipulation for a good chart is that observers can read it accurately. One requirement for good managers, on the other hand, is that they demand the information that they need. As our little test shows, charts based on areas and good management do not go hand in hand. Therefore, when we don’t need anything else to understand management information, we prove that our (visual) judgement is good.
The original chart, which was posted in the Sunday edition of the New York Times (2008–09–21), does contain the actual numbers. I think that this chart serves its purpose…namely, as decoration. Since the New York Times usually serves as a role model for information design, we can deal with a bit of ornamentation on a Sunday. On workdays, however, managers both with and without good visual judgment always should go strictly by the book.
I guessed –55 %, –30 %, +15 %, and –95 %, which is pretty close, but I squinted and figured the percentage of one linear dimension, then squared it in my head. Certainly much harder than estimating the length of bars in a bar chart.
Tuesday, November 11th, 2008, 10:34 pm-55, –36, +21, –96
I used a similar method to Jon, looking for rough fractions on a linear basis and then counting squares added/removed if the whole was divided up on this basis
So the first one looked like thirds, so I imagined in divided into 9, and then it’s clear that one row plus one column minus one for the overlap are missing = 3+2 = 5/9
Similarly 9/25 (a bit high – 11/36 would have been a much better judgement) then 21/100, then 24/25
Using fractions with low denominators makes percentage calculations easy as long as you can remember quick percent _approximations_ and then just multiply by the numerator:
4 25
5 20 (so far so easy, huh?)
6 16.6666 (memory prod: ends with same digit a lot of times)
7 14 (memory prod: 14 is twice seven)
8 12.5 (so use 8 as the converse for 12 or 13)
9 11
So far that list includes four squares: 4,9,16 (near enough for our purpose) and 25
49 is close enough to 50, so I just need to memorise a good number for 36, 64 and 81 and it should be easier in future:
100/36 ~= 2.7 (27 is next lowest multiple of nine, maybe that helps me remember?)
100/64 ~=1.6 (16 is a quarter of 64 and another square)
100/81 =1.23456790123… but 1.25 might be easier to calculate with
This chop-up-by-squares method works (roughly speaking) but is way too much time spent thinking instead of just “seeing” the data
Wednesday, November 19th, 2008, 10:12 amI did pretty much the same thing as the gentlemen above. I was out by 14 % on average–the Berkshire Hathaway guess threw me off because I was trying to estimate the lengths out of 10, and the fraction was a bit smaller than that I guess. But this just goes to show that length comparison is much more effective than area comparison. Unless perhaps we’re comparing geographical or other areas.
Monday, November 24th, 2008, 12:49 amTo be honest, I did not even try:-) Dr. Bizzantz, I guess I am not going to be drawn into this trick
) I believe trying to represent a single number using a chart (% change in market value in this case) is never a good idea. Taking Amin’s point further, 2D is better than 3D and linear better than 2D, if you possibly can help it.
I have a strong hunch that NY Times does these once in a while just so that they can balance out all the other good ones they do all the time
I’ve had a look at another one where they compare the annualized return from the S&P under the democrats and republicans. The chart would end up taking an entire A4 sheet if printed while 1/4th of that would’ve sufficed. Which also shows that even simple innocent 2D charts can be ‘junked-up’.
On a Sunday, on a lazy afternoon surfing session, after a heavy meal….may be…..but I definitely won’t stand such visual debauchery if an important decision were to be made on the basis of that information.
Saturday, November 29th, 2008, 6:09 am