Perhaps you heard that Bill Gates gave a TED talk last week. Here is one of his slides:
How’d he do? Too big? Too small? Just right? If he’s wrong, can you redesign it?
[h/t Steve Phelps, via e-mail]
Also: [WCYDWT] Obama Botches SOTU Infographic, Stock Market Reels
gasstationwithoutpumpsMarch 8, 2011 - 10:13 am -
I measure the diameters as 10 pixels and 324 pixels, which makes an area ratio of 1050, but the desired ratio was 20,822.
Of course, these sorts of graphics are pretty useless, and the black background makes it hard to see the sizes of the shapes when projected anyway so the whole slide is rather pointless.
Ryan BavettaMarch 8, 2011 - 10:58 am -
Using gasstationwithoutpumps’ numbers if you find the ratio of spherical volumes you get a ratio of 34,000:1, which is closer to the 21,000:1.
I wouldn’t expect to compare volumes from the look of the chart though – I would expect a comparison of areas.
NicoMarch 8, 2011 - 11:04 am -
My keen grade 8 students just took the last 15 minutes to run some numbers.
1) The ratio should be 20,822, as confirmed above.
2) Area of the small circle was measured at 0.070685834 cm cubed
3) Area of large circle should be: 1471.852 cm cubed.
4) Actual area of large circle: 246.7401 cm cubed
5) It is off by a factor of 5.965062
Based on radius1= 0.15cm
and radius2= 5cm
Nathan CashionMarch 8, 2011 - 1:08 pm -
I don’t know how gasstation is measuring, but I measure the larger circle at 314 px max.
Nevertheless, using a circle with a diameter of 10 pixels would result in an area of (10/2)^2*3.14 = 78.5 px sq.
The ratio is 62,466,000,000 / 3,000,000 = 20,822:1
Thus, the larger circle should have an area of 78.5 * 20,822 = 1,634,527 px sq.
Divide 1,634,527 by 3.14 = 520286 which we take the square root of to get the radius = 721 pixels or a diameter of 1442.
Unfortunately, Keynote’s default slides only go 1920 pixels high, which results in a slide that looks like these:
Using a black background may not be best in classroom use where lights are often up, but it is the best practice at TED where the lighting is set up appropriately. But I changed the small circle to a white-gray gradient to make it more visible.
(Haven’t really done geometry for years, so let me know if I need to make any corrections.)
Jan van HulzenMarch 9, 2011 - 5:02 am -
For more examples of this i recommend the book How to Lie with Statistics by Darrell Huff.
Another source of errors of this kind is cartoons. The road runner series is the worst. I regularly ask my 4 year old son if he thinks this is possible.
gasstationwithoutpumpsMarch 9, 2011 - 8:16 am -
Nico, how did your students get units of cm^3 for area?
Nathan, you are correct. The number should be 314 pixels, not 324 pixels, for area ratios of 986 and volume ratios of 31,000.
Wesley CalvertMarch 9, 2011 - 1:16 pm -
On the other hand, one must consider the rhetoric of the thing. The end result was off by an area factor of about 20 — how many pixels would that be?
And it’s much better rhetorically to understate the difference by even a large factor than to overstate it by even a little. If you do the former, you get called out on math ed blogs. If you do the latter, it’s the New York Times.
Matt JensenMarch 9, 2011 - 2:32 pm -
I think a nice solution is to use isometric cubes. A small green cube, representing a pile of $3 million, and a large green cube, 27.5 unit cubes in each dimension, representing the 27.5^3-times bigger liability. Make the large cube look like it’s built out of the smaller $3 million cubes.
This could be done with a small cube roughly 11 pixels high, and the large cube roughly 303 pixels high. It doesn’t distort the data, and it’s still easy to read.
Nathan ShieldsMarch 9, 2011 - 10:25 pm -
Here’s a sweet example for the collection. Posted on the Freakonomics blog today: http://www.freakonomicsmedia.com/2011/03/07/the-unintended-consequences-of-polio-eradication-in-india/.
Dan LMarch 10, 2011 - 7:51 am -
I did the Obama SOTU address problem with my class. We were discussing communication and it seemed relevant to point out that what you show could be wrongly interpreted by the listener/viewer.
In the subsequent discussion folks in about equal parts argued: 1) that diameter of a circle is an OK way to show the data and 2) that area should have been used accurately.
I have to say that, as the instructor, I disagree with point 1 in favor of point two. This, leads me to my question:
If you see a circle, what do you think of first? It’s area? Diameter? Equivalent spherical volume?
I think human nature needs to be understood when viewing something like this. Only then can we pick the right representation.
What about LINES for linear comparisons? Rectangles for area comparisons? And shoeboxes for volumetric comparisons? The “anisotropy” of the dimensions may force us to see the relevant metric (volume, area, length).
Nathan CashionMarch 10, 2011 - 10:37 am -
Great point, Dan, about the anisotropy (where did you ever find that word?!). I think your respective use of lines, rectangles, boxes, and Matt’s suggestion to use cubes, is right on.
I would argue that at first look, circles would represent area. Unless the design of it hints towards a sphere. My example above had a shaded gradient on the circle which might make it look more like a sphere, so that is something I would remove to make it more accurate.
JeredMarch 10, 2011 - 11:26 am -
It depends on if these are circles or spheres, of course.
If circles, the RADIUS of the larger circle should be ~144 times the radius of the smaller circle.
Since it was only about 32 times the radius, it actually represents an AREA that is about 20 times too small.
If it’s spheres, then the larger one is a bit too large, as the desired ratio is about 1:27.5
Steve PhelpsMarch 10, 2011 - 11:41 am -
There is another graph in his talk that caused my kids fits. It was the rectangular graph early in the talk. This led me to the same questions you posed. I wondered if the data presented in the rectangular graph was presented in a pie chart instead, would the kids still misread it?
JDLMarch 10, 2011 - 11:43 am -
Here’s another for your collection straight from the midwest. Not sure if it’s mathematically correct or not.
Nathan CashionMarch 11, 2011 - 8:41 am -
I noticed the rectangular graphs that compares Available Resources ($76 Billion) to Budget Shortfall ($25 Billion), a ratio of 3.04:1.
I measured the rectangles of having respective areas of 561,267 pixels and 140,450 pixels, which gives a ratio of 4:1. At first glance it appears the ratios are incorrect.
However, I because the rectangles are overlayed, I wondered if I was looking at it wrong. If $76 Billion + $25 Billion is the total resources needed, then each rectangle represents their respective portion of those resources (those available and those lacking). In this case, the comparison would be ($76 Billion + $25 Billion) / $25 Billion which does give a ratio of 4:1.
Did your students find something different?
JoshMarch 11, 2011 - 10:27 am -
It is great to see so much deep thinking about these graphs. I happened to watch the TED talk last week. I did think a bit about the graphical representation of all the money. However, I also think there is another message here. We are all so concerned about teaching math, about reaching all the kids, and presenting problems that are real and have a hook. And I think we have all gained a great deal by this forum. But I think it almost ironic (well maybe totally ironic) we put all this energy into dissecting the graphs in this talk.
We cut nearly three weeks from the school year because of budget shortfall. The curriculum did not get cut. Class sizes skyrocketed. When(if) things turn around, we will not be compensated for the heaps of work and stress we have endured.
Maybe I am in the wrong profession (is it a profession? I feel like a sparring partner).
Dan MeyerMarch 11, 2011 - 12:54 pm -
Granted. It seems there’s no shortage of places to have the policy debate, though.
Yeah, my sympathies. Seriously. Just as often as I find myself missing the classroom work, I think to myself, “Man, what a mess.”
Steve PhelpsMarch 11, 2011 - 4:23 pm -
My kids thought the entire rectangle represented the 76 Billion, thinking that the 25 Billion should have been one-third, not one-fourth, of the rectangle.
Dan LMarch 12, 2011 - 6:16 pm -
OK, wouldn’t you know that 10 minutes after reading this blog entry, I was looking at an article that committed the same crime to geometry? The only saving grace is that the article scales the x-axis. This leaves the viewer with visual dissonance, but, an easily quantifiable comparison.