David Lockhart

@caltamura3 @instrumenthull I like Long + Ervin (2000) Using heteroskedastic consistent estimators in linear regression jslsoc.sitehost.iu.edu/files_research…

@instrumenthull Could you link to a peer-reviewed article that elaborates on this? Just tryna educate myself

Apropos of nothing, here is a problem set question from one of my undergrad econometrics courses

Here's a technique I just learned for creating random-feeling bijective functions between finite sets of the same size—like for assigning tasks to people. I will use it the next time I have to randomly assign students to groups. • Put the names of students along the top in

An antidote for: If you've got too many people coming to your door, saying, "A quick stats question; it'll only take a minute" ...

In case you ever wonder what REALLY goes on at bachelor parties…

@lathropa It is not my area but I was taught that parents cannot correctly classify recorded cries of their child (gold standard is what action calmed the crying) and that parents typically use context cues (eg long time since feeding, probably hungry) not info in the actual cries

📢New article alert!📢 Have a look at this impressive scoping review by @louisepoppe_ et al on how to develop causal DAGs within health psychology research, including valuable guidance and recommendations on creating DAGs! Read the full article here: doi.org/njzz

Step aside Monty Hall, Blackwell’s N=2 case for the secretary problem is way weirder.

Experimental studies keep giving us more and more evidence against the theory that people have two concepts of causation (dependence vs. production) This is a nice window into how recent work in cognitive science of causal judgment is going beyond previous approaches A thread

everybody asks *what* is probability. Nobody asks *how* is probability

Don’t want to be that guy but this is technically a “causes of effects” question and thus the relevant ATE is not well defined

😳

@RandVegan You might find the zoo package helpful

anyone done alot of time intelligence functions in R? eg. year over year, week over week, month over month, Current ytd vs. prior year ytd, etc? lubricate is really helpful but not sure if there are other packages that also help here

I just discovered something called accidental baroque which are photographs that feel like baroque paintings. I love this as a genre so much. Send me your accidental baroques.

#Today in 1982, in a posting made to a Carnegie Mellon bulletin board, Professor Scott Fahlman proposes the first known use of emoticons While they became famous in 1980s & beyond, their origin remained unknown until 2002, when the original message was retrieved from backup tape

“when a measure becomes a target, it ceases to be a good measure”

@IsabellaGhement @ChelseaParlett Thanks for the explanation. I agree with a lot of it, but to paraphrase Bill James "the alternative to good CI is bad CI not no CI". It is hard to do right but CI theories clarify how to tell the good from the bad. But there's no magic wand that tada makes a param causal.

@avicenna @ChelseaParlett Just because you think upfront about something, it doesn’t mean you are going to get it “right”. That’s ultimately what it boils down to. We all know human thinking is flawed at the best of times. 🤭

Why isn’t EVERYONE doing causal inference??? But also, how is ANYONE doing causal inference???? 😅

@IsabellaGhement @ChelseaParlett Could you say more about that, Isabella?

@ChelseaParlett I may get heat over this, but my own feeling is that causal inference will one day be frowned upon as much as stepwise model selection is frowned upon today. 🫣

@colin_fraser It just like me fr

What if it actually looks like this?

Do you want to use causal graphs to improve your research but never know where to start? In a series of papers, we used the same data and research question—one where we are pretty darn sure we know the real causal effect is that there isn’t one—to compare graph drawing methods.