Stephen Wild
@stephenjwild.bsky.social
I try to put straight lines through things but usually fail. Try to be Bayesian when I can. Views my own. RT/like != endorsement.
John Wick 2 is a great movie. That's what this is about, right? RIGHT?
November 29, 2025 at 1:00 AM
John Wick 2 is a great movie. That's what this is about, right? RIGHT?
And that's when you'll find out that things did not converge
November 28, 2025 at 11:41 PM
And that's when you'll find out that things did not converge
@tslumley.bsky.social, does svyglm()use robust standard errors?
November 28, 2025 at 10:09 PM
@tslumley.bsky.social, does svyglm()use robust standard errors?
I noticed that too. I don't think svyglm() uses robust standard errors, so it's interesting to see them similar.
November 28, 2025 at 9:56 PM
I noticed that too. I don't think svyglm() uses robust standard errors, so it's interesting to see them similar.
Ahh. I missed that you did it with the bottom one. My bad.
November 28, 2025 at 9:51 PM
Ahh. I missed that you did it with the bottom one. My bad.
Got a chance to test my intuition, and if you use svyglm(), the standard errors are about twice the size. Fun.
Interesting article. Good stuff!
Interesting article. Good stuff!
November 28, 2025 at 9:43 PM
Got a chance to test my intuition, and if you use svyglm(), the standard errors are about twice the size. Fun.
Interesting article. Good stuff!
Interesting article. Good stuff!
Though I suspect this is why a bunch of the weights are effectively 0
November 28, 2025 at 5:42 PM
Though I suspect this is why a bunch of the weights are effectively 0
My condolences 😞
November 28, 2025 at 3:26 PM
My condolences 😞
I thought he was named after the cocktail. Today I learned...
November 28, 2025 at 3:10 PM
I thought he was named after the cocktail. Today I learned...
Ready, Set, Bunch!
econometrics of bunching estimators
blogs.worldbank.org
November 28, 2025 at 2:59 PM
October Sky was good.
November 28, 2025 at 1:13 AM
October Sky was good.
Red Dawn is a classic Patrick Swayze movie. It is, how does one say, less than realistic.
November 28, 2025 at 1:12 AM
Red Dawn is a classic Patrick Swayze movie. It is, how does one say, less than realistic.
Red Dawn, if by navigating life you mean fighting the Soviets and other communist countries?
November 28, 2025 at 1:08 AM
Red Dawn, if by navigating life you mean fighting the Soviets and other communist countries?
All I can think of right now is Seinfeld and man hands. iykyk
November 27, 2025 at 11:29 PM
All I can think of right now is Seinfeld and man hands. iykyk
Fine to assume they are gaussian.
Client says "Look at all the changes, they are all negative!" but my point is "if you have a few subjects with very LONG followup, they could drag out the difference since they are in the tails".
I think I'm going to stick with ANCOVA
Client says "Look at all the changes, they are all negative!" but my point is "if you have a few subjects with very LONG followup, they could drag out the difference since they are in the tails".
I think I'm going to stick with ANCOVA
November 27, 2025 at 9:37 PM
If it is, then I would do something like lm(post ~ pre + duration * treatment).
November 27, 2025 at 9:32 PM
If it is, then I would do something like lm(post ~ pre + duration * treatment).
ANCOVA. I don't think you need to include duration unless it's causally related to your outcome
November 27, 2025 at 9:26 PM
ANCOVA. I don't think you need to include duration unless it's causally related to your outcome