This comes directly via John Burk who says he’ll blog about this lesson. But, until then, here’s some of the materials and links that he sent me:
At the moment I’m struggling to figure out exactly what the hook is with this data. I try to avoid “see if you can create a model that predicts what you already know” style questions with data. I think the question I would prefer is “What will the winning time be in 2030?” Then we can have a discussion about the limitation of a model, aided by this article that John also sent my way.
If that’s the question, though, then I might prefer to give them more data. At the same time, maybe this is enough, and it would be kind of cool to compare what the model predicts to what actually happened in 2012.
Maybe the lesson goes something like this: Play two videos at the same time. A video of Bob Hayes winning the 1964 medal, and then Usain Bolt in 2012. The point of this would be to set the context, that people are doing this faster than they used to. Once that context is set, we can ask the question: how fast will folks do this in 2050? We can ask kids what information they need. I’ll have the above data ready. (Why not have the data through 2012 for them? Maybe I’ll just be coy and say that it’ll take too much time to use more data, and they can feel free to add more data points as they need them.)
What else will they need? Some nudging to keep track of this data in a table, which activates all their knowledge about fitting rules to tables. Some graph paper to keep track of all this. Some hints that this is something that they might find helpful. I’ll bet a bunch of them try to calculate the differences.directly on the table and make predictions that way.
But this is a great data set for early work in linear regression.