Baker selected the possible causes and put the data into a regression model to determine the effect each had on his sleep duration. The model showed the strongest effect when he slept with his girlfriend, so he presented her with the data.
I concluded by explaining that, due to her sleep-retardant properties, I could not continue to sleep with her, an act she termed “breaking up”. I should mention that Hermina suggested that my data, being from an observational study rather than an experimental study, only shows correlations rather than causation, and that it was quite possible that I had only chosen to sleep at her apartment on nights when I was less tired, or that I had actually chosen to get less sleep on nights when I had come to her apartment.
She proposed that, instead of taking hasty action, we conduct an experimental study where we flip a coin each night to determine whether I would sleep at her apartment or my own, in order to prove a causative effect. Obviously, I rejected this suggestion. Although this study is insufficient to conclusively prove Hermina’s causative role, this strong a correlation, and the importance of getting enough sleep, are sufficient together to suggest that action needs to be taken expeditiously.
Null Hypothesis is an anarchic and consistently funny UK science magazine that often contains gems like this, as well as curious news from the world of science.