Policy #evaluation inherits concepts from multiple disciplines, and somewhere along the way many of them have become muddled. This blog post sketches what I think are seven of the most irritating myths. andifugard.info/seven-persis...

here’s a free version: neat_research_panel <- data.frame( id = 1:100, treatment = sample(c(TRUE, FALSE), 100, replace = TRUE), outcome = rnorm(100, mean = 50, sd = 5) )

Completed Day 8 of #AdventOfCode in #RStats ‐ fav so far. Skipped part 2 of Day 7. Let's see if I have time for any more! 'Tis busy season! github.com/InductiveSte...

The central place of the bogeyman ‘positivism’ in how social scientists caricature each other’s work is nicely captured here by @adrianblau.bsky.social www.cambridge.org/core/journal...

Whether or not the concerns expressed in this paper are legitimate, the way ‘positivism’ became somehow both meaningless and pejorative is not good.

Day 6 #AdventOfCode was a lot easier #RStats I'm curious to see whether/how people avoided loops. github.com/InductiveSte...

Day 5 #AdventOfCode done in #RStats Used {gmp} when I probably only wanted options(scipen=999). Ah well 🙃 github.com/InductiveSte...

Didn't have much time this eve but Day 4 #AdventOfCode is complete, if slow (10 mins to run). I think the trick is to know when brute-ish force is good enough and just move onto the next day. #RStats github.com/InductiveSte...

Lunchtime fun: I've completed "Lobby" - Day 3 - Advent of Code 2025 #AdventOfCode in #RStats, with the help of recursion and the {gmp} package for big integers. Good fun 🤓 Answers to date here: github.com/InductiveSte...

Logistic regression with 5 million observations. glm() - about 2 hours glm(start = <values found on a tiny subset>) - less than 2 minutes. Almost feels too good... #rstats

Reminder that when the Times had similar worries about Reform, its response was to give Farage this advice on implementation: "Removing a city’s worth of people would be an extended exercise. Identity cards would be a must if the population was to be sieved into legitimate and illegitimate camps"

I'm doing #AdventOfCode again - until I run out of time - using #RStats, of course. Aiming for O(finishes by Christmas day) worst case time complexity. First two solutions (and beyond, if/when I do them) yonder: github.com/InductiveSte...

For the avoidance of doubt: does svyglm() use robust standard errors? Yes! notstatschat.rbind.io/2025/12/01/d...

#rstats It is with profound sadness I heard that my long-time friend and colleague, John Fox passed away this week. He was the author of {car}, {effects}, {Rcmdr}, ... and numerous influential books. I will miss him greatly. www.john-fox.ca

This year explaining regression to the mean took forever but I’m sure next year will be better.

99% of all subgroup analyses give the others a bad name.