Follow-up on Data Representation

posted Aug 29, 2009, 4:58 AM by Donald Vescio   [ updated Sep 12, 2011, 8:10 AM ]
I posted a variant of yesterday's entry on data representation in an active discussion forum.  It generated a good amount of dialogue--over eighty posts in less than a day.  I just added the following commentary that summarizes why I posted the article in the first place:

Great citation--I'm a huge fan of Tufte and have any new members of my staff read this specific text when they arrive on the job.  

My point in my original post was not to dispute the absolute data presented in the original chart, but rather point out how data presentation can significantly impact our reception of it.  By zooming in on the data range, differences look more significant in relative terms.  Of course, if one understands how data works, then the relative component of data reception is (should be) minimal.

Absolute values are very important to me.  I'm competitive enough where a couple of seconds will matter and--despite the manner of presentation of Cervelo's marketing materials--would assume that I'll most likely be riding a P4C for UCI events next season (only caveat is how UCI rules evolve between now and then).

Most consumers, though, don't understand data.  My wife's a triathlon coach and her athletes are extremely successful; what we see regularly are individuals who will read a chart like the original and focus on the relative presentation of data, and not on the absolute values.  This, in turn, leads to a whole series of non-rational decisions, such as choosing a long, low frame for its drag numbers, when a shorter, taller frame really is all that will fit appropriately.

All other variables being equal, a rider will go faster for the same effort/output on a new uber-generation frame than on an older, less aerodynamically efficient model.  But that said, while I still believe that the absolute values that separate framesets are important (and for some riders exceptionally so), the frame has to factor into an analysis of the overall frame/equipment/rider dynamic.  Would a properly designed frame that's short and tall, and which is aerodynamically efficient, perform better than the frame that tests slightly better aerodynamically, but has two inches of round spacers sitting on top of its carefully sculpted head tube?  I suspect that this *could* be the case under some conditions.

The other issue that bothers me about the original chart is that *most* riders will see higher yaw angles than those for which the frames are optimally designed.  Almost any bike is going to be fast at 0 yaw; it's at the larger yaw angles that the most interesting data can be gained.

I fit very well on a Cervelo P3C and have had great success with it.  I also ride fast enough where I'll experience relatively low yaw angles most of the time.  What this tells me is that the P4C would be a great match for me, but also that the absolute differences between this frame and the other frames that one might use are not as great as what might be experienced by an average middle of the pack rider. 

So--the original post had a lot to do with perception and marketing, and not so much with absolute performance.  Even in this thread, its interesting to see how posters fall into three distinct groups--which isn't a bad thing.