Monday, November 9, 2009

The PICA code

I'm a member of the IS Board of Advisors at the University of Missouri St. Louis.  One of the things that this university's IS program struggles with (as do most) is how to graduate better analysts.  I've always seen strong analytical skills as a personality trait rather than something that can be learned.  As I recall my training as a consultant, though, it occurs to me that there were several lessons about how to be a good analyst, how to ask open ended interview questions, how to perform experiments and data analysis to validate assumptions, etc.

The PICA code was a great example of poor analysis.  The project was to add a new field to the pharmacy claim table.  We went and talked with the team that was adding this field to the claim processing system to understand how it was really going to be used and what it really was.  The answer we got was "it's called the PICA code because it can have only one of 4 values... P, I, C, and A.  There was no harm done when we discovered that there were also values of N in that column, though there were lots of jokes about needing the change the column name every time a new value was discovered.

In this case, the weak analysis didn't cause much harm, but there are certainly other cases where poor root assumptions can lead to poor decision making downstream.  I hope that other universities catch up with what UMSL hopes to do in graduating students who are stronger analysts.

Word Count: 14,015

No comments:

Post a Comment