Fischhoff (1975) replication by Seda et al. counted twice in flora.csv (dissertation and preprint of the same study)

Read the source sheet or flora-validation records for FLORA-002628 and FLORA-000314, then inspect R/preprint_dedup.R to understand why the dissertation row is not merged. Keep the dissertation DOI as an alternative identifier on the preprint record, and include the missing OSF v1 and v4 identifiers there. Regenerate the outputs and verify Fischhoff (1975) has two rows in both flora.csv and flora_filtered.csv, with one failed and one successful study.

Summary

FLoRA counts one replication of Fischhoff (1975) twice: Seda, Fatori & Batistuzzo's preprint and the master's dissertation it reports. Both describe the same preregistered study of the same sample (n = 431, Brazil). The study's author pointed this out in May 2026.

• Already fixed: OSF preprint versions v2 and v3 are now merged into one record (FLORA-000314). The merge decision was confirmed in cache/confirmed_preprint_duplicates.csv on 2026-05-05 (commit 99f6f632). Git history cannot show when that merge first reached the published output, because the outputs were restored on 2026-10-23.
• Still wrong: the dissertation is a separate record (FLORA-002628) with doi_r = NA. As a result, flora.csv lists 3 attempts for Fischhoff (1975), 2 of them failed, where there are 2 independent studies. FLoRA Explorer, the replication atlas and flora-pubpeer all read flora.csv.
• Not affected: flora_filtered.csv, used by the API and flora-preprint-notifier, shows 2 attempts. That is correct, but only because filter 1 (invalid_identifier) drops the dissertation row for having no DOI.
• Trap: the dissertation does have a DOI (10.11606/D.5.2024.tde-01082025-155126). If routine identifier cleanup fills it in as a standalone doi_r, the duplicate also enters flora_filtered.csv, the API and the notifications.

The other replication the author mentioned, Chen et al. (2021, JESP), is already in FLoRA and correctly linked. It needs no change.

Data revision checked

• forrtproject/fred-data origin/main: output/flora.csv at commit 4008e783 (2026-10-28 03:06 UTC), 2,818 rows; output/flora_filtered.csv at commit b48a7d61 (2026-10-28 03:09 UTC), 2,400 rows.
• forrtproject/flora-validation: output/flora.csv at commit b968b0b (nightly release 2026-10-28), 2,847 rows. This file has the permanent IDs.
• External records checked on 2026-10-29.

Records for doi_o = 10.1037/0096-1523.1.3.288 (Fischhoff, 1975)

FLORA-001905 (FRED-000281) links the same Chen et al. paper to Slovic & Fischhoff (1977), doi_o = 10.1037/0096-1523.3.4.544, for its Study 2. This is a correct, separate pair.

Reproduce:

git show origin/main:output/flora.csv | Rscript -e 'd <- readr::read_csv(file("stdin"), show_col_types = FALSE); print(dplyr::count(dplyr::filter(d, doi_o == "10.1037/0096-1523.1.3.288"), doi_r, outcome))'
git show origin/main:output/flora_filter_log.csv | grep 1523.1.3.288
git show origin/main:cache/confirmed_preprint_duplicates.csv | grep x394e

Evidence that the records are one study

• Dissertation: Pinto, L. R. S. (2024). Replicação e extensão de Fischhoff (1975): viés de retrospectiva com mediadores clínicos de sintomas depressivos [Master's dissertation, Faculdade de Medicina, Universidade de São Paulo]. https://doi.org/10.11606/D.5.2024.tde-01082025-155126 (registered with Crossref, type dissertation; repository page: https://www.teses.usp.br/teses/disponiveis/5/5142/tde-01082025-155126/). The defence date is 2024-12-16. The supervisor is Batistuzzo and the co-supervisor Fatori, the preprint's co-authors. OpenAlex holds two works for it: W4412991291 (Portuguese title, with the DOI) and W7120759706 (English title, no DOI). FLoRA links the second.
• Preprint: Seda, L., Fatori, D., & Batistuzzo, M. C. (2025). Hindsight Bias: A Preregistered Replication Study of Fischhoff (1975) With an Extension on Depressive Symptoms as Moderation. OSF Preprints. Crossref records four versions linked by is-version-of / has-version:
  • x394e_v1, 2025-03-19
  • x394e_v2, 2025-09-08
  • x394e_v3, 2025-09-10
  • x394e_v4, 2026-03-06; the OSF API marks it is_latest_version.
• Same study: the English abstract of the dissertation and the abstract of preprint v3/v4 report the same design and results almost word for word: a preregistered direct replication of Fischhoff's Experiment 2 in a Brazilian sample (n = 431), no significant difference between retrospective and prospective judgements, and exploratory d = 0.12, 95% CI [0.07, 0.37].
• No journal version found as of 2026-10-29: OSF article_doi is empty for v3 and v4, and a Crossref search by title and author returns only the four preprint versions and the dissertation. Related: #120.
• The other replication is Chen, J., Kwan, L. C., Ma, L. Y., Choi, H. Y., Lo, Y. C., Au, S. Y., Tsang, C. H., Cheng, B. L., & Feldman, G. (2021). Retrospective and prospective hindsight bias: Replications and extensions of Fischhoff (1975) and Slovic and Fischhoff (1977). Journal of Experimental Social Psychology, 96, 104154. https://doi.org/10.1016/j.jesp.2021.104154. The author gave ScienceDirect PII S0022103121000573, which Crossref lists as this DOI's alternative-id. It is a different study (Study 1, N = 890, replicating Experiment 2), coded successful. It is already present (FLORA-001254) and needs no change.

Suggested correction

1. Merge the dissertation into the preprint record. Retire or merge FLORA-002628 (REPL-000334) into FLORA-000314 (REPL-000337), and add 10.11606/D.5.2024.tde-01082025-155126 to alt_identifier_r of FLORA-000314. The preprint dedup in R/preprint_dedup.R can't do this: it groups rows on (doi_o, doi_r) and leaves rows with NA doi_r untouched. The fix therefore needs to be a source-sheet or flora-validation record correction, not an entry in confirmed_preprint_duplicates.csv.
2. Don't fill the dissertation DOI into its own row's doi_r (see Trap above).
3. Add the missing versions 10.31219/osf.io/x394e_v1 and 10.31219/osf.io/x394e_v4 to alt_identifier_r of FLORA-000314. Consider making v4, the latest version, the primary doi_r.
4. Expected result for Fischhoff (1975): 2 rows in both flora.csv and flora_filtered.csv (Seda et al., failed; Chen et al., successful).

Related: #125 (alternative identifiers), #119 ("also published as" linking), #147 (thesis as a venue type), #120 (published DOIs for OSF preprints). None of these covers this case.

Wider check (optional)

Other pairs where a no-DOI thesis or dissertation and a preprint or article by the same author share a doi_o may be counted twice in the same way. Filter 1 hides them from the API, but they stay in flora.csv.