I had wondered about whether the number of students tested by grade might give some indication of enrollment patterns. Perhaps it does, but it looks like it tells us more about patterns of testing. And maybe looking at all the grades for each year is the wrong way to look at this though – in a subsequent post I’ll look at “cohorts” through the system, which could be more interesting. Still, eyeballing the top panel of Figure 5a there is some evidence for a drop between grades 5 and 6, I think. The bottom panel is like Figure 4a with each year exploded out by grade.
It’s a little interesting that in half the years we have data for there was more testing in grade 5 than in grade 4, and in half the years there was more testing in grade 8 than in grade 7. What’s going on there? Again, perhaps looking at cohorts will be illuminating.
Here’s the equivalent graph showing total tests rather than the estimate of students tested. (Here the bottom panel is a closer look at Figure 4b.) Another thing to check out in the future is the Math and ELA counts directly.
One more comment on these lower panel graphs: it might not be really appropriate to connect the points with lines. (See code.) I did it to make it easier to track the adjacent grades. Another approach would be to remove the connecting lines and reinforce the underlying grid, making it more of a Cleveland dot plot.