Rachel Leah Childers
@donskerclass.bsky.social
Econometrics, Statistics, Computational Economics, etc
University of Zurich
http://donskerclass.github.io
🇺🇲 in 🇨🇭. 🏳️⚧️
University of Zurich
http://donskerclass.github.io
🇺🇲 in 🇨🇭. 🏳️⚧️
I'm sympathetic to people with continuous treatments who are mildly put off by average derivative functionals and prefer a conditional line of best fit, the main use case I can see for partially linear models, but it would be good to see an average dose-response curve plotted with it.
November 14, 2025 at 10:34 PM
I'm sympathetic to people with continuous treatments who are mildly put off by average derivative functionals and prefer a conditional line of best fit, the main use case I can see for partially linear models, but it would be good to see an average dose-response curve plotted with it.
Reposted by Rachel Leah Childers
FWIW, there are data-based covariate selection procedures that allow for the possibility that some variables are mediators, not confounders. I even have one, though by now there are quite a few. All still require assumptions, and sometimes you get partial ID, but we can do better than ignoring it.
Local Causal Discovery for Estimating Causal Effects
Even when the causal graph underlying our data is unknown, we can use observational data to narrow down the possible values that an average treatment effect (ATE) can take by (1) identifying the graph...
arxiv.org
November 14, 2025 at 10:17 PM
FWIW, there are data-based covariate selection procedures that allow for the possibility that some variables are mediators, not confounders. I even have one, though by now there are quite a few. All still require assumptions, and sometimes you get partial ID, but we can do better than ignoring it.
Reposted by Rachel Leah Childers
Is it true that I have completely lost my battle to make Double Machine Learning mean "cross-fitting Neyman orthogonal moments with an ML first stage" and not "a partially linear model fit by that procedure"?
People should often be doing the former, useful thing, rarely the latter.
People should often be doing the former, useful thing, rarely the latter.
November 14, 2025 at 10:12 PM
Is it true that I have completely lost my battle to make Double Machine Learning mean "cross-fitting Neyman orthogonal moments with an ML first stage" and not "a partially linear model fit by that procedure"?
People should often be doing the former, useful thing, rarely the latter.
People should often be doing the former, useful thing, rarely the latter.
😡😤🤬
For dozens of different estimators, yes you can fit them by DML in a single line of code. But a different line for each, and many more could be coded in the future. I would like, when I introduce a procedure that works with DML, for the reader to think "any of those". But that's not how it goes.
For dozens of different estimators, yes you can fit them by DML in a single line of code. But a different line for each, and many more could be coded in the future. I would like, when I introduce a procedure that works with DML, for the reader to think "any of those". But that's not how it goes.
November 14, 2025 at 10:23 PM
😡😤🤬
For dozens of different estimators, yes you can fit them by DML in a single line of code. But a different line for each, and many more could be coded in the future. I would like, when I introduce a procedure that works with DML, for the reader to think "any of those". But that's not how it goes.
For dozens of different estimators, yes you can fit them by DML in a single line of code. But a different line for each, and many more could be coded in the future. I would like, when I introduce a procedure that works with DML, for the reader to think "any of those". But that's not how it goes.
FWIW, there are data-based covariate selection procedures that allow for the possibility that some variables are mediators, not confounders. I even have one, though by now there are quite a few. All still require assumptions, and sometimes you get partial ID, but we can do better than ignoring it.
Local Causal Discovery for Estimating Causal Effects
Even when the causal graph underlying our data is unknown, we can use observational data to narrow down the possible values that an average treatment effect (ATE) can take by (1) identifying the graph...
arxiv.org
November 14, 2025 at 10:17 PM
FWIW, there are data-based covariate selection procedures that allow for the possibility that some variables are mediators, not confounders. I even have one, though by now there are quite a few. All still require assumptions, and sometimes you get partial ID, but we can do better than ignoring it.
Is it true that I have completely lost my battle to make Double Machine Learning mean "cross-fitting Neyman orthogonal moments with an ML first stage" and not "a partially linear model fit by that procedure"?
People should often be doing the former, useful thing, rarely the latter.
People should often be doing the former, useful thing, rarely the latter.
November 14, 2025 at 10:12 PM
Is it true that I have completely lost my battle to make Double Machine Learning mean "cross-fitting Neyman orthogonal moments with an ML first stage" and not "a partially linear model fit by that procedure"?
People should often be doing the former, useful thing, rarely the latter.
People should often be doing the former, useful thing, rarely the latter.
"Earnest" is the word I heard more than any other to describe the students at Bowdoin, and it is apt, though I didn't fully understand in what sense it was meant until I was there. Anyway, I think those kids get a good education, and are well placed to live in a world a little better than our own.
November 8, 2025 at 6:33 PM
"Earnest" is the word I heard more than any other to describe the students at Bowdoin, and it is apt, though I didn't fully understand in what sense it was meant until I was there. Anyway, I think those kids get a good education, and are well placed to live in a world a little better than our own.
Congress is certainly a place where deliberate campaign of intimidation of trans people has pushed many of us out, with detriment to the community and the institution. It's certainly not the only such place or the most violent, but it is targeted specifically for its broader impact.
Re BLS news, a few years ago I resigned from a CBO position before starting because I worried that regardless of what I did, the presence of someone like me would be used to impugn the credibility of a highly professional, nonpartisan agency. I can't know, but I can't say I made the wrong decision.
November 7, 2025 at 2:35 PM
Congress is certainly a place where deliberate campaign of intimidation of trans people has pushed many of us out, with detriment to the community and the institution. It's certainly not the only such place or the most violent, but it is targeted specifically for its broader impact.
I have no expertise, but I recently saw Shan Ge from NYU present a paper based on that data, and it seemed quite impressive. This and other papers of hers might have things to say about the properties of the data set: papers.ssrn.com/sol3/papers....
Rising Insurance Premiums Increase Mortgage Delinquency and Drive Relocation to Safer Areas
As climate change intensifies natural disasters, homeowners' insurance premiums are rising dramatically. Using microdata on insurance policies linked to househo
papers.ssrn.com
November 6, 2025 at 10:36 PM
I have no expertise, but I recently saw Shan Ge from NYU present a paper based on that data, and it seemed quite impressive. This and other papers of hers might have things to say about the properties of the data set: papers.ssrn.com/sol3/papers....
Being surrounded by it on all sides made her latest album really click for for me. The opening track:
Ethel Cain - Janie (Official Visualizer)
YouTube video by Ethel Cain
youtu.be
November 4, 2025 at 10:53 PM
Being surrounded by it on all sides made her latest album really click for for me. The opening track:
This is just the study protocol, not the full study with results, which as far as I can tell has not yet been released, beyond the press summary.
October 30, 2025 at 11:05 PM
This is just the study protocol, not the full study with results, which as far as I can tell has not yet been released, beyond the press summary.
Wait... I was at that con at the time. Probably not in the same line unless it was for one of the larger events that most participants attended. I have no desire to reclaim or even use the term, but I do think it would be funny to grant dispensations on that basis.
October 29, 2025 at 10:56 PM
Wait... I was at that con at the time. Probably not in the same line unless it was for one of the larger events that most participants attended. I have no desire to reclaim or even use the term, but I do think it would be funny to grant dispensations on that basis.
That's a new one for me! I spent a summer on the Upper West Side, and took numerous day trips when I lived in CT, but I never made it to that one.
October 22, 2025 at 6:26 PM
That's a new one for me! I spent a summer on the Upper West Side, and took numerous day trips when I lived in CT, but I never made it to that one.
That's a strong yes. Thanks!
October 22, 2025 at 6:11 PM
That's a strong yes. Thanks!
Congrats on snagging tickets to Oh, Mary! Tickets are left for over $500 and I at least thought about it.
Cats is a little less up my alley. (Sorry)
I'll probably just go to an art museum before meeting a friend at NYU for dinner, then spend Sunday shopping.
Cats is a little less up my alley. (Sorry)
I'll probably just go to an art museum before meeting a friend at NYU for dinner, then spend Sunday shopping.
October 22, 2025 at 6:08 PM
Congrats on snagging tickets to Oh, Mary! Tickets are left for over $500 and I at least thought about it.
Cats is a little less up my alley. (Sorry)
I'll probably just go to an art museum before meeting a friend at NYU for dinner, then spend Sunday shopping.
Cats is a little less up my alley. (Sorry)
I'll probably just go to an art museum before meeting a friend at NYU for dinner, then spend Sunday shopping.
Clearly the middle of the sandwich should be F for Falafel Information Matrix.
October 20, 2025 at 5:56 PM
Clearly the middle of the sandwich should be F for Falafel Information Matrix.
Pardon me, it was color corrector, not foundation. Also, some of those 2x2 matrices had incorrect transposes. And I of course ate shit and hurt my tailbone on the skateboard. It is through these mistakes that we learn...
October 20, 2025 at 1:33 PM
Pardon me, it was color corrector, not foundation. Also, some of those 2x2 matrices had incorrect transposes. And I of course ate shit and hurt my tailbone on the skateboard. It is through these mistakes that we learn...
Suddenly realized 20+ years later that S., the slightly older Australian lesbian in our Final Fantasy 7 IRC chat who wrote a long fic about a boy who woke up as a girl and then got to go on adventures and had a great time and learned a lot, maybe might have not been cisgender.
Hope she's doing well.
Hope she's doing well.
October 18, 2025 at 10:10 PM
Suddenly realized 20+ years later that S., the slightly older Australian lesbian in our Final Fantasy 7 IRC chat who wrote a long fic about a boy who woke up as a girl and then got to go on adventures and had a great time and learned a lot, maybe might have not been cisgender.
Hope she's doing well.
Hope she's doing well.