#ResearchIntegrity #PNAS #OpenScience #Reproducibility #DataFabrication #AcademicTwitter #PoliticalScience #SocialScience #RobustnessChecks #ModelUncertainty.

(6/6). 🧠Bottom line:.Fraud makes headlines. But our study shows even honest research can flip depending on design. Robustness isn’t about one decision — it’s about the entire modeling universe. 📄Paper: 🧪Replication:

(5/6). So what drives this fragility?. ❌Not omitted controls. ✅YES to:.• How the sample is selected.• How the outcome is defined. 👇Animated plot of what matters most:. 🟢= significant positive. 🔴= significant negative. 🔵= not significant

(4/6). 🚨Key result:.The same variable can be significantly positive or negative — depending on how you specify the model. This isn’t fraud. It’s structural fragility. 👇Animated plot of result flipping:. 🔴= significant negative. 🔵= significant positive

(3/6). We applied this to four core topics in political science:. 🏛️Democratization. 💸Welfare generosity. 📊Public goods provision. 🧠Institutional trust. Across all four, we analysed a large number of defensible model — leading to over 3.6 billion estimates in total.

(2/6). Most robustness checks tweak a few control variables. But what if you:. 📦Change the sample?. 🎯Redefine the outcome?. 🧱Switch fixed effects?. 🧮Use different standard errors?. We combined multiverse analysis + Extreme Bounds Analysis to test it all at once.

🚨 New paper out in PNAS @PNASNews 🚨. Following recent high-profile research fraud scandals, we asked:. 👉 How stable are findings—even when researchers follow the rules?. With Tim Vlandas (Oxford), we analyzed 3.6B regression coefficients. Surprising results ahead. 🧵⬇️. (1/6).

10/. 📄Read the full article: Would love to hear from others working on campaigns, voter behavior, or political psychology!. #PoliticalScience #Elections #CampaignPromises.

9/.So yes — economic promises can win votes. But the effect is fleeting. They change behavior, not identity.

8/.These voters didn’t suddenly agree with the CDU on everything. They just responded to a targeted promise that addressed their needs.

7/. 💡Implications:. 🔹Campaign promises can have real, causal effects. 🔹They’re especially powerful for opposition voters facing economic insecurity. 🔹But the effect fades once the benefit is delivered.

6/.In short:. ➡️The CDU gained voters because of the promise. ➡️But lost them again after fulfilling it.A temporary alignment, not a permanent conversion

5/. 📊Key findings:. ✅Support for the CDU jumped 12.7%among eligible mothers. ✅The biggest gains came from low-income, left-leaning voters. 🚫No effect on politically disengaged citizens. 📉And once the benefit was delivered — the effect disappeared

4/.Because of the cutoff, I could isolate the causal effect of the promise — separate from delivery, ideology, or party loyalty. So what happened?.

3/.To find out, I studied a rare natural experiment from Germany. In 2013, the CDU promised a generous “Mother’s Pension” — but only for moms whose kids were born before January 1992. This arbitrary cutoff created a clean test.

2/.Big promises often define elections:. 🧱Trump’s border wall. 👶Brexit’s £350m/week for the NHS. 🎓Biden’s student debt relief. But do they actually change how people vote?.

1/. 🚨New solo-authored paper out now in Comparative Political Studies:. “From Pledge to Poll: Investigating the Impact of Campaign Promises on Party Alignment”. 📄 Published in:

🧵 Did Trump’s wall actually win him votes — or was it just political theater?. What if a campaign promise alone could sway voters — even before it’s delivered?. 👇.

RT @IMFNews: Some governments faced massive public protests as they tried to implement policies to meet the 2015 Paris climate agreement. N….

RT @DSPI_Oxford: Are policies designed to avert climate change politically costly? . Davide Furceri, @GanslmeierM, Jonathan D. Ostry find t….