Alex Coppock
@aecoppock.bsky.social
Associate Professor of Political Science at Northwestern University
alexandercoppock.com
Persuasion in Parallel: https://alexandercoppock.com/coppock_2023.html
Research Design: Declaration, Diagnosis, and Redesign: book.declaredesign.org
alexandercoppock.com
Persuasion in Parallel: https://alexandercoppock.com/coppock_2023.html
Research Design: Declaration, Diagnosis, and Redesign: book.declaredesign.org
This paper is great up through table 4 -- very nice RCT evidence that free gym increases exercise and course completion.
At table 5, they start controlling for post-treatment variables and drawing unsupported conclusions about mechanisms.
Remove to improve!
At table 5, they start controlling for post-treatment variables and drawing unsupported conclusions about mechanisms.
Remove to improve!
University students who were provided with a free gym card (in a randomized experiment) exercised more and had a significant improvement in academic performance. The treated students were also less likely to drop out of classes
www.journals.uchicago.edu/doi/full/10....
www.journals.uchicago.edu/doi/full/10....
University of Chicago Press Journals: Cookie absent
www.journals.uchicago.edu
November 24, 2025 at 9:47 PM
This paper is great up through table 4 -- very nice RCT evidence that free gym increases exercise and course completion.
At table 5, they start controlling for post-treatment variables and drawing unsupported conclusions about mechanisms.
Remove to improve!
At table 5, they start controlling for post-treatment variables and drawing unsupported conclusions about mechanisms.
Remove to improve!
Here's hoping for dozens and dozens of experimental estimates of the effects of anti-corruption messages on vote choice.
We now have enough studies of populist messages for a meta-analysis (papers.ssrn.com/sol3/papers....); maybe soon we'll be able to do the same for anti-corruption?
We now have enough studies of populist messages for a meta-analysis (papers.ssrn.com/sol3/papers....); maybe soon we'll be able to do the same for anti-corruption?
November 20, 2025 at 8:15 PM
Here's hoping for dozens and dozens of experimental estimates of the effects of anti-corruption messages on vote choice.
We now have enough studies of populist messages for a meta-analysis (papers.ssrn.com/sol3/papers....); maybe soon we'll be able to do the same for anti-corruption?
We now have enough studies of populist messages for a meta-analysis (papers.ssrn.com/sol3/papers....); maybe soon we'll be able to do the same for anti-corruption?
(prohibition) persuasion in parallel
Had fun reconstructing the data from this 90-year old persuasion experiment that shows that "wet" and "dry" college students update their attitudes in the direction of counterattitudinal persuasive information.
paper: doi.org/10.1080/0022...
Had fun reconstructing the data from this 90-year old persuasion experiment that shows that "wet" and "dry" college students update their attitudes in the direction of counterattitudinal persuasive information.
paper: doi.org/10.1080/0022...
November 19, 2025 at 10:16 PM
(prohibition) persuasion in parallel
Had fun reconstructing the data from this 90-year old persuasion experiment that shows that "wet" and "dry" college students update their attitudes in the direction of counterattitudinal persuasive information.
paper: doi.org/10.1080/0022...
Had fun reconstructing the data from this 90-year old persuasion experiment that shows that "wet" and "dry" college students update their attitudes in the direction of counterattitudinal persuasive information.
paper: doi.org/10.1080/0022...
Audit experimenters of Bluesky... did you know there's a creator on TikTok who is calling churches, synagogues, and mosques asking if they have baby formula for her two month old?
This is a "one condition" study (no causal inference) but hoo boy the descriptives...
www.tiktok.com/@nikalie.mon...
This is a "one condition" study (no causal inference) but hoo boy the descriptives...
www.tiktok.com/@nikalie.mon...
Islamic Center of Charlotte in Charlotte NC. Would help feed a starving baby no hesitation 🥰🥰🥰 #fyp #fypシ #fypシ゚viral #testingyourchurch #church #faith #religion #baby #hungrybabytest #viral #viralvid...
TikTok video by Nikalie 🌈
www.tiktok.com
November 13, 2025 at 3:23 PM
Audit experimenters of Bluesky... did you know there's a creator on TikTok who is calling churches, synagogues, and mosques asking if they have baby formula for her two month old?
This is a "one condition" study (no causal inference) but hoo boy the descriptives...
www.tiktok.com/@nikalie.mon...
This is a "one condition" study (no causal inference) but hoo boy the descriptives...
www.tiktok.com/@nikalie.mon...
a “Space Force”, if you will
we need a department of kerning
November 13, 2025 at 3:17 PM
I'm one of those academics who measure message persuasiveness via survey experiment. I think it's absolutely the right method for doing so.
I think the response to this good point from @anatosaurus.bsky.social is not to abandon survey exps but instead to measure "getting heard" (attention) also.
I think the response to this good point from @anatosaurus.bsky.social is not to abandon survey exps but instead to measure "getting heard" (attention) also.
November 5, 2025 at 6:13 PM
I'm one of those academics who measure message persuasiveness via survey experiment. I think it's absolutely the right method for doing so.
I think the response to this good point from @anatosaurus.bsky.social is not to abandon survey exps but instead to measure "getting heard" (attention) also.
I think the response to this good point from @anatosaurus.bsky.social is not to abandon survey exps but instead to measure "getting heard" (attention) also.
lol at the MPSA shade in the last line of the FAQ
November 5, 2025 at 4:52 PM
lol at the MPSA shade in the last line of the FAQ
Maybe your texts tell you to "BE A VOTER" like mine.
It all started because of a PNAS paper that claimed that the noun form it increased voter turnout (relative to the verb form ) by 11 to 14 percentage points.
It keeps not replicating, obviously.
Most recently doi.org/10.1017/bpp....
It all started because of a PNAS paper that claimed that the noun form it increased voter turnout (relative to the verb form ) by 11 to 14 percentage points.
It keeps not replicating, obviously.
Most recently doi.org/10.1017/bpp....
November 4, 2025 at 6:25 PM
Maybe your texts tell you to "BE A VOTER" like mine.
It all started because of a PNAS paper that claimed that the noun form it increased voter turnout (relative to the verb form ) by 11 to 14 percentage points.
It keeps not replicating, obviously.
Most recently doi.org/10.1017/bpp....
It all started because of a PNAS paper that claimed that the noun form it increased voter turnout (relative to the verb form ) by 11 to 14 percentage points.
It keeps not replicating, obviously.
Most recently doi.org/10.1017/bpp....
Doing meta-reanalysis means my coauthors and I are constantly asking for data.
This month we heard back from 21 of 32 author teams on first email, all responses "yes" or "soon." (nice!)
Innovation: asking authors to post data publicly rather than privately with us via email (4 did, more will!)
This month we heard back from 21 of 32 author teams on first email, all responses "yes" or "soon." (nice!)
Innovation: asking authors to post data publicly rather than privately with us via email (4 did, more will!)
October 30, 2025 at 4:26 PM
Doing meta-reanalysis means my coauthors and I are constantly asking for data.
This month we heard back from 21 of 32 author teams on first email, all responses "yes" or "soon." (nice!)
Innovation: asking authors to post data publicly rather than privately with us via email (4 did, more will!)
This month we heard back from 21 of 32 author teams on first email, all responses "yes" or "soon." (nice!)
Innovation: asking authors to post data publicly rather than privately with us via email (4 did, more will!)
chat's personalized flattery is getting out of hand
October 29, 2025 at 3:46 PM
chat's personalized flattery is getting out of hand
rarely have we needed the "suggestive" results from a "factors associated with" design to have the possibility that X affects Y "suggested" to us
This is an excellent point that generalizes.
Researchers often defend suboptimal practices by referring to future studies with better designs.
But: Why would anybody run those studies when you can just throw a bunch of variables into a regression and make sweeping "preliminary" claims?
Researchers often defend suboptimal practices by referring to future studies with better designs.
But: Why would anybody run those studies when you can just throw a bunch of variables into a regression and make sweeping "preliminary" claims?
October 28, 2025 at 12:29 PM
rarely have we needed the "suggestive" results from a "factors associated with" design to have the possibility that X affects Y "suggested" to us
encountered this article by David Weimer from 1986 (cited by 7) with three good ideas for improving science that just a few short decades later, we've started to implement.
Collective Delusion In The Social Sciences: Publishing Incentives For Empirical Abuse
doi.org/10.1111/j.15...
Collective Delusion In The Social Sciences: Publishing Incentives For Empirical Abuse
doi.org/10.1111/j.15...
October 27, 2025 at 12:28 PM
encountered this article by David Weimer from 1986 (cited by 7) with three good ideas for improving science that just a few short decades later, we've started to implement.
Collective Delusion In The Social Sciences: Publishing Incentives For Empirical Abuse
doi.org/10.1111/j.15...
Collective Delusion In The Social Sciences: Publishing Incentives For Empirical Abuse
doi.org/10.1111/j.15...
Reposted by Alex Coppock
@chloethurston.bsky.social & Mary McGrath (both at @polisciatnu.bsky.social) join the Perspectives on Politics editorial team as Associate Editors in American Politics!
Chloe Thurston and Mary McGrath join the Perspectives on Politics editorial team as Associate Editors
APSA is happy to announce that Chloe Thurston, Associate Professor of Political Science and Mary McGrath are joining the Perspectives on Politics editorial team as Associate Editors in American…
APSA is happy to announce that Chloe Thurston, Associate Professor of Political Science and Mary McGrath are joining the Perspectives on Politics editorial team as Associate Editors in American…
Chloe Thurston and Mary McGrath join the Perspectives on Politics editorial team as Associate Editors
APSA is happy to announce that Chloe Thurston, Associate Professor of Political Science and Mary McGrath are joining the Perspectives on Politics editorial team as Associate Editors in American Politics. Chloe Thurston, Associate Professor of Political Science; Faculty Fellow, Institute for Policy Research at Northwestern University Professor Thurston’s research is on American political development, political economy, and public policy, with a particular interest in how politics and public policy shape market inequality, including along the lines of race and gender.
politicalsciencenow.com
October 20, 2025 at 3:53 PM
@chloethurston.bsky.social & Mary McGrath (both at @polisciatnu.bsky.social) join the Perspectives on Politics editorial team as Associate Editors in American Politics!
Congrats to @daniel-a-n-goldstein.com and @drewstommes.bsky.social for showing how an information treatment that US wars are fought predominantly by the poor increases preferences for recruiting from the rich.
No effect on tax policy tho! here again we see fx on "target attitudes" not "nontarget"
No effect on tax policy tho! here again we see fx on "target attitudes" not "nontarget"
Do citizens want to address unfair burden-sharing in public goods? New research out in ISQ @isq-jrnl.bsky.social with @drewstommes.bsky.social addresses this question by highlighting the unequal socioeconomic costs of US wars
academic.oup.com/isq/article/...
academic.oup.com/isq/article/...
October 17, 2025 at 4:24 PM
Congrats to @daniel-a-n-goldstein.com and @drewstommes.bsky.social for showing how an information treatment that US wars are fought predominantly by the poor increases preferences for recruiting from the rich.
No effect on tax policy tho! here again we see fx on "target attitudes" not "nontarget"
No effect on tax policy tho! here again we see fx on "target attitudes" not "nontarget"
What are you doing Thursday Nov 13? This panel on survey methods for measuring "rare traits" seems v. important especially if we think small numbers of people sometimes have outsize influence on politics.
The speakers appear to be Bluesky-less, but I'll tag the moderator @mattgraham.bsky.social!
The speakers appear to be Bluesky-less, but I'll tag the moderator @mattgraham.bsky.social!
Rare traits like support for violence, conspiracy beliefs, and unsafe health behaviors are really hard to measure! On Thursday November 13, learn from experts about the problem and solutions.
Free for students and AAPOR members, $5 for others. Sign up here: www.eventbrite.com/e/the-rare-t...
Free for students and AAPOR members, $5 for others. Sign up here: www.eventbrite.com/e/the-rare-t...
October 16, 2025 at 3:41 PM
What are you doing Thursday Nov 13? This panel on survey methods for measuring "rare traits" seems v. important especially if we think small numbers of people sometimes have outsize influence on politics.
The speakers appear to be Bluesky-less, but I'll tag the moderator @mattgraham.bsky.social!
The speakers appear to be Bluesky-less, but I'll tag the moderator @mattgraham.bsky.social!
👀 this new meta-analysis on edutainment by @bardiarahmani.bsky.social, Montano, @dylanwgroves.bsky.social, and Green
doi.org/10.1017/bpp....
377 ests in 77 exps: edutainment moves attitudes, norms, beliefs 📊
Effects persist ⏳
Many reasonable theories about effect heterogeneity are not supported 😇
doi.org/10.1017/bpp....
377 ests in 77 exps: edutainment moves attitudes, norms, beliefs 📊
Effects persist ⏳
Many reasonable theories about effect heterogeneity are not supported 😇
October 14, 2025 at 5:40 PM
👀 this new meta-analysis on edutainment by @bardiarahmani.bsky.social, Montano, @dylanwgroves.bsky.social, and Green
doi.org/10.1017/bpp....
377 ests in 77 exps: edutainment moves attitudes, norms, beliefs 📊
Effects persist ⏳
Many reasonable theories about effect heterogeneity are not supported 😇
doi.org/10.1017/bpp....
377 ests in 77 exps: edutainment moves attitudes, norms, beliefs 📊
Effects persist ⏳
Many reasonable theories about effect heterogeneity are not supported 😇
Reposted by Alex Coppock
Currently in FirstView: In “Balancing Precision and Retention in Experimental Design,” @gustavodiaz.org and Erin Rossiter study how experimental design choices can increase precision when estimating treatment effects. Specifically, they examine block-randomized and pre-post designs.
October 14, 2025 at 4:35 PM
Currently in FirstView: In “Balancing Precision and Retention in Experimental Design,” @gustavodiaz.org and Erin Rossiter study how experimental design choices can increase precision when estimating treatment effects. Specifically, they examine block-randomized and pre-post designs.
Reposted by Alex Coppock
Since it’s the #equinox, time to re-post the best #dataviz I’ve ever done: 365 pictures of the Earth from space, taken at exactly 6:00 a.m. Details on the NASA Earth Observatory: earthobservatory.nasa.gov/images/52248...
September 22, 2025 at 9:06 PM
Since it’s the #equinox, time to re-post the best #dataviz I’ve ever done: 365 pictures of the Earth from space, taken at exactly 6:00 a.m. Details on the NASA Earth Observatory: earthobservatory.nasa.gov/images/52248...
love this #notamap as a way of understanding the scale of land use. we're so used to seeing state borders and we have a rough sense of their relative sizes -- this plot makes great use of that
Below is a clever visualization of the distribution of how land is used in the US. Cows roam over a lot of land!
But this is not a map - for example, the 100 largest landowning families aren't confined to Florida.
But this is not a map - for example, the 100 largest landowning families aren't confined to Florida.
September 15, 2025 at 8:53 PM
love this #notamap as a way of understanding the scale of land use. we're so used to seeing state borders and we have a rough sense of their relative sizes -- this plot makes great use of that
The great Vincent Arel-Bundock shows how to use marginaleffects post-estimation summaries with DeclareDesign simulations!
The new {marginaleffects} release for #RStats (0.30.0) comes with two new vignettes:
1. Speed up computation with automatic differentiation (often 10x gains) marginaleffects.com/bonus/perfor...
2. Power analyses with {marginaleffects} and {DeclareDesign}. marginaleffects.com/bonus/power....
1. Speed up computation with automatic differentiation (often 10x gains) marginaleffects.com/bonus/perfor...
2. Power analyses with {marginaleffects} and {DeclareDesign}. marginaleffects.com/bonus/power....
37 Performance – Model to Meaning
marginaleffects.com
September 14, 2025 at 2:52 PM
The great Vincent Arel-Bundock shows how to use marginaleffects post-estimation summaries with DeclareDesign simulations!
Reposted by Alex Coppock
Grateful to Matt Baum, Jamie Druckman, @jonmladd.bsky.social @brendannyhan.bsky.social @dannagal.bsky.social for joining the Author Meets Critics session on The American Mirage tomorrow at 12:00. If you spot me at the conference, ask for my favorite book merch: a custom magnetic bottle opener!
September 12, 2025 at 6:12 PM
Grateful to Matt Baum, Jamie Druckman, @jonmladd.bsky.social @brendannyhan.bsky.social @dannagal.bsky.social for joining the Author Meets Critics session on The American Mirage tomorrow at 12:00. If you spot me at the conference, ask for my favorite book merch: a custom magnetic bottle opener!
I'm once again a session chair for APSA and will be imposing the Nyhan / Samii rules. Encourage your chairs to do the same.
I don't see any good arguments for the usual
all papers
all discussant comments
all questions
format
I don't see any good arguments for the usual
all papers
all discussant comments
all questions
format
I'm a session chair for APSA (Innovations in Experimental Design, 4pm at the Marriott!) and will IMPOSE this format.
90 minutes / 5 papers = 18 minutes a paper
10 minutes presentation [no time for lit review, friends!]
2 minutes discussant [praise + 1 good point]
6 minutes Q&A [2, maybe 3 Qs]
90 minutes / 5 papers = 18 minutes a paper
10 minutes presentation [no time for lit review, friends!]
2 minutes discussant [praise + 1 good point]
6 minutes Q&A [2, maybe 3 Qs]
For folks going to APSA - friends don't let friends use the horrible default panel format. Divide the time equally and open the floor discussion *after each talk*. Works SO much better as long as the chair enforces time limits. cyrussamii.com?p=1806
September 4, 2025 at 9:24 PM
I'm once again a session chair for APSA and will be imposing the Nyhan / Samii rules. Encourage your chairs to do the same.
I don't see any good arguments for the usual
all papers
all discussant comments
all questions
format
I don't see any good arguments for the usual
all papers
all discussant comments
all questions
format
whaaaat!? that's so cool!
literally just:
weird_trick = D * X
iv_robust(weird_trick ~ D | Z)
to get the mean of X among compliers!
literally just:
weird_trick = D * X
iv_robust(weird_trick ~ D | Z)
to get the mean of X among compliers!
September 3, 2025 at 7:27 PM
whaaaat!? that's so cool!
literally just:
weird_trick = D * X
iv_robust(weird_trick ~ D | Z)
to get the mean of X among compliers!
literally just:
weird_trick = D * X
iv_robust(weird_trick ~ D | Z)
to get the mean of X among compliers!
Reposted by Alex Coppock
All statistical models are wrong, but some are stupid
September 3, 2025 at 1:35 PM
All statistical models are wrong, but some are stupid
Suppose there are only two publication tracks for an experiment:
1) Pre-registered experiment:
PAP -> Realization -> Submit -> RR -> Publication
2) Registered report:
PAP -> Submit -> RR -> Realization -> RR2 -> Publication
when should we want 2 not 1?
to protect against "null risk"?
1) Pre-registered experiment:
PAP -> Realization -> Submit -> RR -> Publication
2) Registered report:
PAP -> Submit -> RR -> Realization -> RR2 -> Publication
when should we want 2 not 1?
to protect against "null risk"?
What's the issue with registered reports? They seem like a good idea on the surface, which is the depth of my knowledge.
September 3, 2025 at 3:26 PM
Suppose there are only two publication tracks for an experiment:
1) Pre-registered experiment:
PAP -> Realization -> Submit -> RR -> Publication
2) Registered report:
PAP -> Submit -> RR -> Realization -> RR2 -> Publication
when should we want 2 not 1?
to protect against "null risk"?
1) Pre-registered experiment:
PAP -> Realization -> Submit -> RR -> Publication
2) Registered report:
PAP -> Submit -> RR -> Realization -> RR2 -> Publication
when should we want 2 not 1?
to protect against "null risk"?