Partly Cloudy, 64°
Earlier today, a statistics professor was reportedly struggling to establish causality between a bloody knife and his gaping stab wound.
“You’d think this would be a pretty clear case of causality, but really all we know is that there is a strong positive correlation between our variables,” said Professor Peter Briston, wincing as he pulled the handle of the blood-soaked knife from his abdomen. “Let’s think about what other factors might have contributed to this bleeding from my side. What are some possible sources of omitted variable bias?
“Consider the counterfactual: If I wasn’t assigned the knife, our treatment, would we still observe this bleeding outcome?” asked Briston, slurring his words and stumbling about as more of his blood pooled at his feet. “This is a perfect situation to open up Stata and run a difference-in-difference regression.”
“Oh, that’s, uh, not great. Yeah, that doesn’t help us,” continued Briston, who could be referring to either the results of the regression or the large quantities of blood continuing to gush out of his midsection. “So our beta-one coefficient doesn’t seem to be statistically significant at the 95% confidence level. I—I think I’m going to go sit down and think on this one.”
At press time, paramedics uncovered a note in Briston’s pocket proving causality between Tech House membership and celibacy.
