One of Our Favorite Regression Lessons
If teaching statistics is a meal, then regression is the desert. There are so many compelling, real-world examples that we can explore with regression. Plus, regression analyses can lead to rich discussions on correlation vs. causation.
Still, in the past, we fell into the trap of picking boring datasets - just because they made for nice linear fits. Plus, we found ourselves teaching correlation vs. causation in entertaining but contrived terms (e.g. per capita consumption of margarine and divorce rates). Don't get us wrong: seeing good linear models is important, and we love introducing correlation/causation with spurious correlations. Yet, there is more we can do to get students thinking critically using regression.
Below, we feature one of our favorite regression lessons. The topic is relatable (schools), important (education policy), and features a very subtle correlation vs. causation insight (why didn't test scores change?). Try it out and let us know how it goes for your class.
Let’s skew it!
Least Squares Regression
Lesson 3.2
Attendance is a strong predictor of test scores. So, is fixing attendance the key to improving test scores? Students use regression to find a surprising result.