Linking research content, NIH funding, and employment broadens scientific workforce evaluation: An application to U.S. drug and alcohol research — Supplement and evidence

This deposit accompanies “Linking research content, NIH funding, and employment broadens scientific workforce evaluation: An application to U.S. drug and alcohol research.” It contains the study evidence and two reproducibility tools as three independently runnable ZIP components in one Zenodo record. 1. study_evidence_v9.zip: the S1–S4 supplement and coding codebook; aggregate population, validation, research-content, funding and employment tables; institutional analysis panel and NIH funding inputs; saved estimates and diagnostics; deposited figures; searchable catalog, dictionaries and provenance notes; and Python numerical verification/table-export code. 2. proquest_v5_reproducibility.zip: the saved ProQuest query, search/export instructions, revision-15 codebook, reader prompts and response schemas, frozen configuration, notebook, processing workflow and synthetic tests for a new authorized ProQuest TDM Studio run. No licensed source records, prebuilt roster, identifier list or credentials are included. A new search may retrieve different records; this is a method-replication route, not exact historical record/model-judgment replay. Authenticated model-snapshot availability was not verified by the offline tests. 3. rc_rmi_scorer.zip: the standalone RC-RMI occupation–industry instrument, recorded component judgments, scales, combination rules, crosswalk, scoring code, synthetic examples and full-domain checks under three specifications. It reproduces frozen rule application, not historical employment matching or individual-duty validation. Reproduction capabilities vary by analysis. Supplied aggregates support arithmetic reconstruction and table export; the institutional panel supports eight fresh count-model fits. Other intervals, model estimates and diagnostics are retained as saved results. Licensed source records/text, identifying linkage files, individual employment evidence, paired coding labels and bootstrap draws are excluded. Start with the record-level README.md, then the README inside the relevant ZIP. CITATION.cff identifies the combined record; component citation files retain the versions of the tools. Reuse terms vary by material: original software uses MIT, original expression and instrument judgments use CC BY 4.0 within the stated scopes, and empirical/third-party materials retain their respective source-use conditions. Consult the component license notices, DATA_AVAILABILITY.md and file/table-level rights metadata.