Andy Timm
@andytimm.bsky.social
Data scientist @ Grow Progress, testing Dem ads. Usually thinking about methods for survey weighting, efficient causal inference, or scalable Bayes. https://andytimm.github.io/
Kevin Munger has both some great descriptive work and meta scientific argument articulating the value of descriptive work for fast moving topics like social media platforms in his book The YouTube Apparatus!
See also his temporal validity paper for an articulation of the metascientific point.
See also his temporal validity paper for an articulation of the metascientific point.
May 29, 2025 at 3:35 PM
Kevin Munger has both some great descriptive work and meta scientific argument articulating the value of descriptive work for fast moving topics like social media platforms in his book The YouTube Apparatus!
See also his temporal validity paper for an articulation of the metascientific point.
See also his temporal validity paper for an articulation of the metascientific point.
Another incredible essay.
Some striking passages to get you interested:
Some striking passages to get you interested:
January 18, 2025 at 2:18 PM
Another incredible essay.
Some striking passages to get you interested:
Some striking passages to get you interested:
It’s happening 😃
(The trailing off of today’s data in the graphs is a bit hard to read, but oh well)
(The trailing off of today’s data in the graphs is a bit hard to read, but oh well)
January 6, 2025 at 1:03 PM
It’s happening 😃
(The trailing off of today’s data in the graphs is a bit hard to read, but oh well)
(The trailing off of today’s data in the graphs is a bit hard to read, but oh well)
We’re internalizing your externalities 😃
January 5, 2025 at 1:22 PM
We’re internalizing your externalities 😃
I just need to make a Zaller one, and then with this and the Converse one below I’ll be able to respond to ~90% of events in politics with the same dumb bit.
January 4, 2025 at 10:04 PM
I just need to make a Zaller one, and then with this and the Converse one below I’ll be able to respond to ~90% of events in politics with the same dumb bit.
Not my best work, but it'll do.
January 4, 2025 at 9:32 PM
Not my best work, but it'll do.
Cause, Effect, and the Structure of the Social World I think? At least as far as more broadly applicable stuff.
www.bu.edu/bulawreview/...
www.bu.edu/bulawreview/...
January 2, 2025 at 2:13 PM
Cause, Effect, and the Structure of the Social World I think? At least as far as more broadly applicable stuff.
www.bu.edu/bulawreview/...
www.bu.edu/bulawreview/...
I don’t entirely agree with Bailey’s pessimistic assessment, but seeing him flesh out Meng’s (2018) framework for survey error more fully was quite helpful.
Many of the most interesting conversations I had about polling this year were sparked by this book or the earlier paper + discussion... 1/2)
Many of the most interesting conversations I had about polling this year were sparked by this book or the earlier paper + discussion... 1/2)
December 30, 2024 at 11:33 PM
I don’t entirely agree with Bailey’s pessimistic assessment, but seeing him flesh out Meng’s (2018) framework for survey error more fully was quite helpful.
Many of the most interesting conversations I had about polling this year were sparked by this book or the earlier paper + discussion... 1/2)
Many of the most interesting conversations I had about polling this year were sparked by this book or the earlier paper + discussion... 1/2)
This is probably my favorite political science book this year.
Hollowness is a fantastic metaphor for problems on both the Democratic and Republican parties in the US, and the book is comprehensive on the history of how “hollowing out” happened, and incisive in describing the harms this causes.
Hollowness is a fantastic metaphor for problems on both the Democratic and Republican parties in the US, and the book is comprehensive on the history of how “hollowing out” happened, and incisive in describing the harms this causes.
December 30, 2024 at 11:33 PM
This is probably my favorite political science book this year.
Hollowness is a fantastic metaphor for problems on both the Democratic and Republican parties in the US, and the book is comprehensive on the history of how “hollowing out” happened, and incisive in describing the harms this causes.
Hollowness is a fantastic metaphor for problems on both the Democratic and Republican parties in the US, and the book is comprehensive on the history of how “hollowing out” happened, and incisive in describing the harms this causes.
Between undergrad and grad school, I learned some basic convex optimization. Then in grad school, I realized I had severely underestimated how useful it was and learned more.
Late last/early this year, as I was learning more about regularized raking/writing rswjax (github.com/andytimm/rsw...) (1/2)
Late last/early this year, as I was learning more about regularized raking/writing rswjax (github.com/andytimm/rsw...) (1/2)
December 30, 2024 at 11:33 PM
Between undergrad and grad school, I learned some basic convex optimization. Then in grad school, I realized I had severely underestimated how useful it was and learned more.
Late last/early this year, as I was learning more about regularized raking/writing rswjax (github.com/andytimm/rsw...) (1/2)
Late last/early this year, as I was learning more about regularized raking/writing rswjax (github.com/andytimm/rsw...) (1/2)
Intro to Modern Causal Inference is probably the top thing I read this year!
This was amazing for learning the “structure” underlying building efficient causal estimators, and helped me unify my understanding of approaches like bias-correction, estimating equations (e.g. DML), and TMLE! (1/2)
This was amazing for learning the “structure” underlying building efficient causal estimators, and helped me unify my understanding of approaches like bias-correction, estimating equations (e.g. DML), and TMLE! (1/2)
December 30, 2024 at 11:33 PM
Intro to Modern Causal Inference is probably the top thing I read this year!
This was amazing for learning the “structure” underlying building efficient causal estimators, and helped me unify my understanding of approaches like bias-correction, estimating equations (e.g. DML), and TMLE! (1/2)
This was amazing for learning the “structure” underlying building efficient causal estimators, and helped me unify my understanding of approaches like bias-correction, estimating equations (e.g. DML), and TMLE! (1/2)
Vibecession remains a bit wild and rather disturbing.
I kinda expected something like this, but seeing the chart is still jarring.
From ( @pkrugman.bsky.social ‘s new substack).
I kinda expected something like this, but seeing the chart is still jarring.
From ( @pkrugman.bsky.social ‘s new substack).
December 21, 2024 at 3:34 PM
Vibecession remains a bit wild and rather disturbing.
I kinda expected something like this, but seeing the chart is still jarring.
From ( @pkrugman.bsky.social ‘s new substack).
I kinda expected something like this, but seeing the chart is still jarring.
From ( @pkrugman.bsky.social ‘s new substack).
Very cool paper addressing some challenges in effect generalization!
December 13, 2024 at 5:24 PM
Very cool paper addressing some challenges in effect generalization!
Converse strikes again
December 13, 2024 at 2:01 PM
Converse strikes again
Ooph, rough one for me (Phil Converse, political scientist famous for work on the average voter lacking real structure and stability in their political opinions).
December 11, 2024 at 11:09 PM
Ooph, rough one for me (Phil Converse, political scientist famous for work on the average voter lacking real structure and stability in their political opinions).
Waffling on which laser eye converse I like more
December 9, 2024 at 10:53 PM
Waffling on which laser eye converse I like more
I especially appreciate this point in footnote 4, re vibes in discussing the LPM.
December 3, 2024 at 6:37 PM
I especially appreciate this point in footnote 4, re vibes in discussing the LPM.
Soft ornaments are the most exciting cat toys!
November 30, 2024 at 5:40 PM
Soft ornaments are the most exciting cat toys!
Candles: lit.
Little xmas tree: up.
Soft ornaments: stolen by Hypnos.
Happy cozy season y’all.
Little xmas tree: up.
Soft ornaments: stolen by Hypnos.
Happy cozy season y’all.
November 30, 2024 at 12:50 AM
Candles: lit.
Little xmas tree: up.
Soft ornaments: stolen by Hypnos.
Happy cozy season y’all.
Little xmas tree: up.
Soft ornaments: stolen by Hypnos.
Happy cozy season y’all.
Very cool little extension of what we can learn from IVs in persuasion experiments.
November 27, 2024 at 5:32 PM
Very cool little extension of what we can learn from IVs in persuasion experiments.
If you don’t get to control the design, use a ROL that correctly handles unknown preferences for middle options. I show both a full likelihood and faster approximation for this in Stan. Both outperform MaxDiff!
(Full ROL here is just for comparison; utilities are normalized)
(Full ROL here is just for comparison; utilities are normalized)
November 25, 2024 at 7:36 PM
If you don’t get to control the design, use a ROL that correctly handles unknown preferences for middle options. I show both a full likelihood and faster approximation for this in Stan. Both outperform MaxDiff!
(Full ROL here is just for comparison; utilities are normalized)
(Full ROL here is just for comparison; utilities are normalized)
If you have sufficient control over designing the experiment, you can use a rank ordered logit on a connected graph of choices.
The connected graph allows you to ask respondents for just best/worst on 3-item sets, but still collect the full ordering for the ROL!
The connected graph allows you to ask respondents for just best/worst on 3-item sets, but still collect the full ordering for the ROL!
November 25, 2024 at 7:36 PM
If you have sufficient control over designing the experiment, you can use a rank ordered logit on a connected graph of choices.
The connected graph allows you to ask respondents for just best/worst on 3-item sets, but still collect the full ordering for the ROL!
The connected graph allows you to ask respondents for just best/worst on 3-item sets, but still collect the full ordering for the ROL!
Converse strikes again
January 16, 2024 at 2:50 PM
Converse strikes again
Someone should make a GPT that is incredibly excited to offer you the latest public domain characters and art criticizing IP law in the US
January 1, 2024 at 4:13 PM
Someone should make a GPT that is incredibly excited to offer you the latest public domain characters and art criticizing IP law in the US