Best Method to Chart Numerous Data Points

[candimc]

I need to develop charts demonstrating health care providors self-
assessment ratings versus average ratings on an individual and group
basis. I also need to display changes over time.

I have 10 participants who evaluated their performance over 303
practices (which are divided across 24 topic areas). The ratings are
on a scale of 1 to 3.

Since there are so many areas evaluated (the 303) I am having
difficulting determining what the best chart methodology to utilize
is. Even if I calculate the average per participant for each of the
24 topic areas, all the charts I come up with are extremely
cluttered.

Any suggestions would be greatly appreciated. Thank you.

 

[Del Cotter]

I need to develop charts demonstrating health care providors self-
assessment ratings versus average ratings on an individual and group
basis. I also need to display changes over time.

I have 10 participants who evaluated their performance over 303
practices (which are divided across 24 topic areas). The ratings are
on a scale of 1 to 3.

Since there are so many areas evaluated (the 303) I am having
difficulting determining what the best chart methodology to utilize
is. Even if I calculate the average per participant for each of the
24 topic areas, all the charts I come up with are extremely
cluttered.

It's hard to know what to suggest without more detail. Do you need a
full time series, or will two times do? I would produce ten small line
graphs in a 2 x 5 array, each showing the average ratings (y axis) for
one participant (one chart) for each topic (x axis) for the current and
previous time period (red and blue line series). The ratings for the
other nine participants for the current period could be there too, in
light grey, for comparison within graph as well as between graphs.

I tend to prefer MEDIAN() as a measure of average instead of AVERAGE(),
which calculates the mean. Instead of all the other participants in a
jungle of grey lines, perhaps you'd prefer a grey median line and light
grey area band of upper and lower quartiles. The red and blue lines
could also be flanked by lighter red and blue quartile lines. So for
each participant you should be able to see a profile of average
performance against the whole set of 303 practices, variation around the
average, and change in average and variation with time.

For more ideas, consider asking on the web sites for "Juice Analytics"
and "Perceptual Edge", which are both discussion forums about the
problem of getting useful actionable performance information from
complex changing multivariate data sets.