Conventions are there for a reason (normally)

This arrived in my inbox recently:
With the helpful commentary:
"The graph shows that for the academic year 2013/14 Y11 students with 100% attendance got As on average. Those with 95% attendance got Cs, those with 90% attendance got Fs and those with 85% attendance got Us."
Where to start?
  1. The nature of the commentary implies a causal relationship between attendance and average points score - that is, some link between a particular attendance and the score. For that to be graphed, we would expect the cause (attendance) to be on the x-axis and the effect (points) to be on the y-axis. This chart is the other way round - points score seems to cause attendance.
  2. Presumably, the line is a linear "best fit" - but it's not labelled -- it is this line ( I presume) that is being used to generate the commentary - but I'm not convinced that parents would "get that".
How is the commentary of "100% attendance got A's on average" actually illustrated by this chart? I guess partents are expected to go up the y-axis to 100%, find the intersection with the best fit line and read the x-axis value of 50. Then to know the conversion between points and grades. Easy? errrr No!!

  • Plot attendance on the x-axis
  • Points don't need to start at 0
  • Too much precision 10.0, 20.0 etc -- display to 0 dp
  • The "best fit" line is anything but linear

Some tweaking (done as an image as I don't have the raw data)
This chart now emphasises the importance of higher attendance. Below about 80% attendance, points score doesn't drop off as fast -- as the majority of the decrease is seen in the 100-85% range. So, attendance (high) does matter.

comments powered by Disqus