Using drawings of cats* as a source of data

*insert your animal of preference here

The first time I tried using a drawing-based interactive in my intro stats class was back in summer school 2018. I put up a stopwatch on the screen, handed out pieces of paper, then asked students to draw a cat and record how long it took them (with a one minute cut off). I collected all the drawings (hundreds of them!) and put them in a box.

So many cat drawings!

[Side note: Being able to use the students in a large lecture as a population and demonstrate random sampling in action is one of the many benefits of teaching large intro stats courses. One of the motivations of using physical drawings of cats was that there was “need” to sample in order to learn about this large population.]

The plan was to use a random sample of the drawings to answer the questions: How many seconds on average did students in our lecture take to draw their cat? Did most students in our lecture draw “heads only”? Before we took a random sample of drawings, I asked students to write down what they thought the answers could be to both questions. I then asked for a couple of student volunteers to help me select the random sample of cat drawings from the box. Each drawing was shown on the document camera live, and like with many live demonstrations things didn’t exactly go to plan!

Below is an excerpt from the lecture recording – remember one of the questions was Did most students in our lecture draw “heads only”?

The next summer when I tried this activity again, instead of asking students to record their drawing time (the Google Quick! Draw! data set is better for investigating this), I asked them to draw a cat on one side of a card, and then turn the card over and rate the lectures in the course so far on a scale of 0 to 100. I’ll admit this was a pretty scary thing to do live but I had a good sense by this point in teaching programme that the ratings would not be bad.

I also asked them to record whether they liked cats or not, so we could compare lecture ratings based on whether a student like cats or not, since well …. cats feature heavily in my lecture materials 😺😺. Similar to the first approach, we used a random sample of cat drawings to make inferences about all drawings/students in the lecture.

A sur-purr-ising result?

And because I had done this cat drawing activity for two semesters, we could compare the drawing tendencies across two different populations!

Still a preference for cat bodies over faces …..

When teaching summer school this year, I used a new app I’ve developed for capturing drawings digitally, this time for an activity to (re)introduce students to different types of variables e.g. numeric vs categorical. Students were asked to draw a bunch of flowers using their devices, I displayed all their drawings (see below) and then asked them to create as many numeric or categorical variables as they could from the drawings.

Some of my student’s drawings of flowers

Using drawings as a source of data is great for building students awareness of what data is or could be. It’s always great to use data about students themselves, and extending this “personal” approach to using drawings they have created allows for new kinds of variables to be explored instead of the classic heights, weights, eye colours etc. It’s also fun and engaging – whether I use the hands-on approach of physical drawings or the digital approach, the participation rate for drawing-based interactives is very high!