A(nother) Statistics and Data Science Blog
This blog is a recording of rumination on various statistical topics that I'm currently thinking about. I hope to have each piece be a more cleaned up and professional version of my thoughts on a couple of papers or books I am reading. Currently, I'm plannning on writing something on Lindley's Paradox and Stein's Paradox because I am familiar with each and it is haunting having this page be so emtpy.
Here are a few statistics blog I regularly consume. I urge you to take a look there before reading anything here. They are bound to be of higher quality, and I have learned more there than I can hope to retain.
Gelman et al.'s Statistical Modeling, Causal Inference, and Social Science .
Simonshon et al.'s Data Colada .
Debroah Mayo's Error Statistics Philosophy .
Richard McElreth's Elements of Evolutionary Anthropology .
Posts
Abstract.
Stein's paradox, introduced by Charles Stein in 1955, uncovers some fascinating insights in statistical decision theory. It shows that when you have a decision rule and a risk function, one estimator can actually be shown to be better than another—though it doesn’t claim to be the best. At first, this might feel a bit disappointing, but it really highlights some important connections in statistics, like the blurry line between inference and decision-making. Plus, it opens the door to modern methods like multilevel models that build on these ideas.
Works in Progress
Abstract.