Avoiding Flame Wars

Before my life as a statistician/data analyst, I was a special education teacher in the Bronx with an undergraduate degree in English with a Spanish minor. At the time, I viewed the worlds of math and science as free from the subjective/emotional debates waged within the liberal arts. Little did I know, that turf wars seem to be endemic to all fields. During my graduate studies in statistics, I discovered that there were even intense disagreements about the fundamental nature of probability.

I have to admit that I enjoy observing these online discussions. However, as with many things on the internet, these discussions sometimes devolve into very personal debates. I look forward to exploring different philosophical ideas within statistics and data science on this blog, but on certain topics, I would prefer to be an observer rather than a participant. To that end, I’ve compiled a list of six statistics/data science topics that tend to stoke conflict on social media.

List of Minefields to Avoid

  1. Bayesian vs. Frequentist Statistics
  2. Linear vs. Logistic Regression for Binary Outcomes
  3. R vs. Python
  4. DAGS/SCM vs. Potential Outcomes for Causal Inference
  5. Poisson Regression with Quasi-MLE vs. Negative Binomial Regression for Count Data
  6. Survey Weighting vs. Multi-level Regression with Post-Stratification for Survey Data

This list is by no means exhaustive, and I may add more topics later.

Let me know if you’ve noticed any statistical topics that seem to provoke emotions online.