Cyrus Samii
@cdsamii.bsky.social
NYU Politics prof. Methods to inform policy. Governance, conflict, institutions. cyrussamii.com
Political science
26%
Mathematics
24%
Pinned
Forthcoming article "Interest-based Negotiation over Natural Resources: Experimental Evidence from Liberia" Darin Christensen, Alexandra C Hartman @cdsamii.bsky.social and Alessandro Toppeta
@eeanews.bsky.social
Teaching materials available: www.eeassoc.org/teaching-mat...
doi.org/10.1093/jeea...
@eeanews.bsky.social
Teaching materials available: www.eeassoc.org/teaching-mat...
doi.org/10.1093/jeea...
Interest-Based Negotiation over Natural Resources: Experimental Evidence from Liberia
Abstract. We experimentally evaluate whether an interest-based negotiation training for community leaders in Liberia improves their ability to strike benef
doi.org
Now in print at JEEA: results from a negotiation training RCT for community leaders in rural Liberia. The goal was to address experience/knowledge asymmetries between community leaders and commercial interests in negotiating mining, logging, land lease, and other resource use contracts.
Facing seasoned commercial negotiators, community leaders may lose out on rents. Interest-based negotiation training improves skills (measured w/ lab-in-the-field methods) and leads to lower rates of communal forestland exploitation, suggesting increased scrutiny of potential forest use deals.
Reposted by Cyrus Samii
Interest-Based Negotiation over Natural Resources: Experimental Evidence from Liberia
Abstract. We experimentally evaluate whether an interest-based negotiation training for community leaders in Liberia improves their ability to strike benef
doi.org
Forthcoming article "Interest-based Negotiation over Natural Resources: Experimental Evidence from Liberia" Darin Christensen, Alexandra C Hartman @cdsamii.bsky.social and Alessandro Toppeta
@eeanews.bsky.social
Teaching materials available: www.eeassoc.org/teaching-mat...
doi.org/10.1093/jeea...
@eeanews.bsky.social
Teaching materials available: www.eeassoc.org/teaching-mat...
doi.org/10.1093/jeea...
Reposted by Cyrus Samii, David Darmofal
The Causal Revolution: Why Research Design trumps (regression) models for causal claims
YouTube video by Alexander Wuttke
youtu.be
How to explain the “credibility revolution” in a few minutes?
"I" "made" a short video that tries to show why clever (natural) experiments and research design beat pure statistical adjustment for causal claims.
I am genuinely curious what methods people think:
youtu.be/Fv14ktwA31Q?...
"I" "made" a short video that tries to show why clever (natural) experiments and research design beat pure statistical adjustment for causal claims.
I am genuinely curious what methods people think:
youtu.be/Fv14ktwA31Q?...
I’ll check it out 👍
www.nobelprize.org
Link to Nobel lecture: www.nobelprize.org/uploads/2018...
@scottfabramson.bsky.social you also mentioned McFadden. Have a look at his Nobel lecture, pp. 334-5. He does a barefoot (unconstrained) comparison of model predictions (*calibrated* to data from a demand survey) to actual outcomes. The model fitting was not a basis for testing the model.
Another thought: let's distinguish model calibration from severe testing (in Mayo's sense). Credibility revolution is a framework for thinking about severity. One can calibrate models on endogenous data, but to test accuracy, use unconstrained credibility revolution methods, a la Todd & Wolpin '06.
I think Ashworth et al. offer a compelling way to think about the relationship between theoretical models and empirics. I’d tell students to start there.
A great example of combining the strengths of various methods is this paper by Allen et al. (a recent favorite of mine):
www.science.org/doi/10.1126/...
www.science.org/doi/10.1126/...
Quantifying the impact of misinformation and vaccine-skeptical content on Facebook
Low uptake of the COVID-19 vaccine in the US has been widely attributed to social media misinformation. To evaluate this claim, we introduce a framework combining lab experiments (total N = 18,725), c...
www.science.org
Agreed: bsky.app/profile/cdsa...
Quantifying the impact of misinformation and vaccine-skeptical content on Facebook
Low uptake of the COVID-19 vaccine in the US has been widely attributed to social media misinformation. To evaluate this claim, we introduce a framework combining lab experiments (total N = 18,725), c...
www.science.org
A great example of combining the strengths of various methods is this paper by Allen et al. (a recent favorite of mine):
www.science.org/doi/10.1126/...
www.science.org/doi/10.1126/...
Survey experiments' popularity in political science is getting attention. What is good and bad about them? How can one maximize their benefits and mitigate their downsides?
Greg Huber and I wrote up our thoughts:
Paywalled: doi.org/10.1016/bs.h...
Free: m-graham.com/papers/Huber...
Greg Huber and I wrote up our thoughts:
Paywalled: doi.org/10.1016/bs.h...
Free: m-graham.com/papers/Huber...
One historical note: the original Dehejia and Wahba pscore paper, and particularly the Smith and Todd comment, showed that incorporating pre-treatment outcomes was crucial (in that case to address the “Ashenfelter dip” mean reversion problem).
Yes I think the Hazlett and Xu paper did a good job in discussing the convergence.
I was reading this Rosenbaum 2010 paper on evidence factors recently and you can see from his example 2 that he had already incorporated a version of this into his non parametric identification logic (matching first differenced data): academic.oup.com/biomet/artic...
Other paper I like that had developed similar ideas:
Hazlett & Xu 2018: yiqingxu.org/packages/tjb...
Imai et al. 2023: onlinelibrary.wiley.com/doi/abs/10.1...
Hazlett & Xu 2018: yiqingxu.org/packages/tjb...
Imai et al. 2023: onlinelibrary.wiley.com/doi/abs/10.1...
arxiv.org
So I guess we are back to matching? (Although, by the references cited, the authors appear to be unaware of the connection?)
arxiv.org/pdf/2511.21977
arxiv.org/pdf/2511.21977
The credibility revolution is more than just a brand. It is an important methodological turn in social science. Social scientists need to understand what it actually involves, both on its own terms and because it now anchors shared standards for evidence.
I also don't think RCTs have a primary purpose to test theories about causal processes that occur naturally. They may be motivated by such theories, but in the first instance the goal is to learn about the impact of an intervention, which may be unlike anything that occurs naturally.
I think that reflects a common misunderstanding of what RCT means. RCT⊂ Randomized experiment, but RCT≠Randomized experiment. The difference comes in the "trial" component, which references a real-world test of an intervention. So, I think your point holds for randomized experiment, but not RCT.
I wouldn't say "similar" actually. Quite different issues (which accepts that the latter is indeed an issue too).
I think the end of the blog post makes this distinction.
A blog post giving a more thorough take on survey experiments and the credibility revolution: cyrussamii.com?p=4168
Despite these database problems, I do find tools like Elicit and Undermind (both not for profit), which use OA and SS as source data, as better search tools than GS. This shows that one could substitute GS by building on OA and SS.
The data quality point is based on my experience — citations are garbled too often and different versions of the same paper are not linked. My presumption is that a messy database contributes to your point 2. Point 1 could be addressed if operating inside an academic consortium with access rights.
Academic institutions should pool resources to make either OpenAlex or Semantic Scholar reliable alternatives to Google Scholar. OA and SS have open architectures with APIs making them more versatile than GS in principle. The problem now is poor data quality; so the task is mainly curating the data.
Yes, your point 2.
New paper! @william-dinneen.bsky.social @guygrossman.bsky.social Yiqing Xu and I use GPT to code 91k articles from 174 polisci journals (2003–2023)and track research designs, transparency practices, and citations. How has the credibility revolution reshaped the discipline? doi.org/10.31235/osf...
🧵
🧵
This is an impressive project. My reaction to what it shows though is that survey experiments have gotten out of hand in polisci. I will blog more on this, but I do not think survey experiments are emblematic of the credibility revolution. Some are already interpreting as such, which is a problem.
New paper! @william-dinneen.bsky.social @guygrossman.bsky.social Yiqing Xu and I use GPT to code 91k articles from 174 polisci journals (2003–2023)and track research designs, transparency practices, and citations. How has the credibility revolution reshaped the discipline? doi.org/10.31235/osf...
🧵
🧵