Graphical lies and semi truths

I’ve been geeking out a little with some data in the last 24 hours, looking at the circulation stats of my blokes for the last 2 years plus the last month.  Glancing through the table leads to the conclusion that they are in fact borrowing more books – I know that doesn’t necessarily lead to more reading, but I also know that just having books around is a good thing.  One of the issues is that inherent in international schools is the fact that students come and go. Most of our students have been here for the last 2 years, and a few entered mid-way – so I averaged the monthly book borrowing by assuming 8 months of school (yup, we’re off for 16 weeks of the year), and then apportioned appropriately. (Note 2016/7 data needs to be updated at end of September to give the full 1.5 months).

But no-one likes looking at a table of data, so how to get this into a graph?  Now even a 3rd grader would (hopefully) be able to tell you that discrete data = bar chart.  So that leads to this:

That kind of shows you the picture – that the green bars are generally the highest (although maybe I need to invert the colours).

Now, let me show you another picture. This time I am quite incorrectly showing you a line graph.  Why is it wrong – well because a line graph is to show the relationship between two sets of values, with one set being dependent on another. Well, as each point is a different child’s reading,  and one child’s data has nothing to do with another child’s data, so obviously a line graph is nonsense. Except for the fact that it much more clearly shows that students have increased their borrowing since they’ve joined BWB. Quite wrong, but more graphically. I’ve been even more deceptive by ordering the data by number of books by date (mainly because the first 6 boys were not at the school in 2014/5 and the next 3 not last year so it made things look more confusing if I didn’t order it.

The next graph is even more pretty but it’s wrong wrong wrong and very deceptive – because I used a “stacked line” it’s no longer showing the boys who they read less than the previous year (s) as I’m adding up their reading over time.

No wonder they say “lie, damned lies and statistic” – maybe they need to add “graphs” to that one.

More problems with this type of data – it tells you about the quantity, but nothing about the quality of what’s being borrowed. If I drill into various circulation histories I see a lot of “churn” of graphic novels. I’m assuming the lads who’ve read 15 or 20 books in the past month are reading nearly one book a day. That’s assuming they’re reading them. In fact one of the boys who seems to be borrowing and reading less is the boy who is tackling much more sophisticated literature and longer books.

Which shows just how individualised one’s approach needs to be to students, and data, and even goals and aims.  I like to think I’m employing a “bait and switch” tactic in the long term – I wonder if that can be quantified? Thinking aloud – if my books were lexiled and I could for each child see a lexile trail that gets stronger over time … unfortunately wishful thinking at this point as our books aren’t lexiled. I’m wondering if any longitudinal research has been done in this respect? I was watching a demo of Scholastic’s Literacy Pro yesterday and maybe that’s influencing my thinking. Because once you’re in that kind of program the program in itself corrupts the data by only feeding the student books in their lexile range, so you have animals in captivity rather than in the wild, if you see what I mean.

Now to see what the reading data tells me, and to see if there is any way to tie it all together in a pretty picture.  I need a spare mathematician to guide me through this – first year university stats is just not going to cut it I fear!