Friday, March 5, 2010

Bad Graph

One of the things I teach when I teach process improvement classes is how to display information.  People taking my classes will lead process improvement projects that are only successful if processes get improved (that is, if people change the way they do things).  A major component of being successful at getting people to change is showing them things, like how bad the current process is, how much it could improve, where the problems are, etc, etc.  Much of this showing will occur graphically.  So I teach them how to be persuasive graphically.  How to distill information and tell a story that is true and compelling - in a picture.

Here is a graph that the Wall Street Journal published yesterday accompanying a story about genetic variations and weight loss.  A few years ago some researchers compared 4 common diets and concluded that there was not much difference in success rates.  Later, some other researchers collected DNA from the 138 study participants and analyzed it for gene variants, assigning the study participants into groups depending on whether their gene variants favored low carb diets or low fat diets.  When they looked at the earlier weight loss in light of the new data on genotype, they found that when people who were predisposed to low carb tried a low carb diet, they lost much more weight than people who tried diets that didn't match their predisposition.

I love this stuff.  It's a geek's dream and low carb vs low fat is a tender and meaningful subject for me.  I also love the detailed treatment Wall Street Journal gave the subject.  The local paper covered it in summary fashion.  I expect many news outlets will touch on this study - it's news!

Back to the graph.  A graph is good when an average reader can understand and correctly interpret it without too much effort.  Readers of the Wall Street Journal are probably smarter than average.  I gotta say though - it took me a good 10 minutes and several attempts at this graph before I could figure it out.  The story it's telling is not that complex, so the graph doesn't need to be this difficult.  Take a look for yourself!

Here's a hint:  there are two axes, on the bottom and top.  All the green circles come from the bottom axis and all the blue circles come down from the top axis.  See if you can take it from there.  (As a graph geek, I'd really love to get at the raw data for this.  I'd like to see the variability and group size represented by each of the circles.)

This slides perfectly into my problem with the main stream media (MSM).  Due to relentless cost-cutting, there's no staff for anyone to do any actual reporting.  Much "news" comes from press releases or new books being published.  So why does this piece of science reporting give me heartburn?  Here's why.  This material is only news because some lab developed a $149 test for the gene markers for low carb vs low fat.  This lab is the group that performed the second study and published the results.  I wonder how much business this is going to drum up for them.  Their potential market is huge.  Anyone who has tried a diet and failed (and can read or watch TV) will now be able to spend $149 and find out whether they're a low carb person or a low fat person.  What a piece of marketing genius!

1 comment: