GCRI · Labs
연구 프로토콜
A research protocol fixes, before data are collected, what a study asks, how it will answer and what would count as an answer: the hypotheses, the design, the controls, the outcomes, the sample and the analysis. Many failures of credibility trace back to a protocol that was vague, unwritten or changed after the data were seen. This page sets out how protocols are built, registered, reviewed and reported, and what ethics committees, editors and funders rely on when they accept a study’s claims.
일이 진행되는 방식.
프로토콜 및 재현 보드
Research protocols · Labs · Quarter 3
프로토콜
| PR-378 sensor drift | 현장 실험 | 예 | 승인됨 | 실행 중 |
| PR-705 AI triage | 무작위화 | 예 | 조건 | 설정 |
| PR-694 message recall | 설문 실험 | 예 | 승인됨 | 완료 |
| PR-360 heat exposure | 코호트 | 아니요 | 대기 중 | 초안 |
Replication R-25
- 원래 효과
- 0.4 SD
- 재현 효과
- 0.2 SD
- 판정
- 더 작고, 같은 부호
- 발행됨
- 데이터와 코드와 함께
테스트베드 가용성
그림Protocol and replication board: example names and illustrative values, not a record or a live system.
Research protocols: how the risk develops.
Hypotheses, controls and scientific acceptance. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
Flexibility becomes bias
When outcomes, exclusions, covariates or stopping rules are chosen after the data are seen, many analyses become possible and the most striking one tends to be reported. Each choice looks reasonable alone; together they inflate false positives and effect sizes.
Weak controls confound results
A missing control, unblinded assessment or non-random allocation lets something other than the intervention explain the difference. Design flaws cannot be repaired by analysis; they travel into every result and review built on the study.
Underpowered designs mislead twice
Small samples miss real effects, and the effects they do detect tend to be overestimated. Underpowered studies that reach significance seed literatures that later, larger replications cannot match.
Deviations go unreported
Protocols change for good reasons, such as slow recruitment or a safety signal. When a change is not dated, explained and reported, readers cannot tell a justified amendment from outcome switching, and reviews pool what should not be pooled.
How research protocols is modelled, measured and tested.
모형, 그 가정, 데이터, 불확실성을 과학자나 엔지니어가 확인할 형태로 제시합니다. 구조는 발행된 방법을 따릅니다. 제시된 값은 예시이며 결과가 아닙니다.
모형, 방정식 및 가정
The model structure used for research protocols, the assumptions it makes and how each is tested.
Ri = f(Hi, Ei, Vi, Ci)재해, 노출, 취약성, 대처 역량EAL = ∑k pk · Lk연간 확률이 pkxt+1 = A xt + B ut + εtA carries the couplings, u the shock, ε the error| 가정 | 만드는 이유 | 시험 방법 |
|---|---|---|
| 한 사건 안에서 재해와 노출은 서로 독립적임 | 손실 적분을 다루기 쉽게 유지함 | 결합 기록에 적합한 코퓰러와 비교 |
| 결합은 운전점 근처에서 선형이다 | 충격이 A를 통해 전파되게 함 | 선형 범위를 넘어 섭동을 주고 비교하십시오 |
| 취약성 곡선은 장소 간에 옮겨 쓸 수 있다 | 지역 곡선은 거의 없다 | 관측된 손실이 있는 곳에서는 이에 대해 검증 |
- 투입 자료
- 출처 정보가 있는 버전 관리 데이터세트
- 산출물
- 분포이며 결코 단일 숫자가 아님
- 코드
- 모델 카드 및 시험과 함께 공개
- 재현
- 같은 데이터, 코드, 환경과 시드
Protocols are registered before results in every domain.
Protocols feed the Labs 증거 시리즈. Registered 보고서 and EQUATOR guidelines set practice; studies with people follow the Declaration of Helsinki (revised 2024).
- Protocols for review.
- Replication proposals.
측정되는 것과 그 정확도.
각 지표에는 단위, 주기, 그 뒤의 출처 종류, 값에 붙을 근거 표시가 있습니다. 값은 표시하지 않습니다. 값은 프로그램이 자체 권한으로 제공합니다.
| 지표 | 단위 | 주기 | 출처 유형 | 라벨 |
|---|---|---|---|---|
| Registration before data collection | percent of studies with a time-stamped registration | 연간 | Trial and study registries | 공표 통계 |
| Primary outcome consistency | percent of reports matching the registered primary outcome | 연간 | Registry and publication comparison | 관측 |
| Sample size justification | percent of protocols with a power or precision rationale | 연간 | Protocol audit | 관측 |
| Reporting guideline adherence | percent of applicable checklist items reported | 연구별 | Assessment against CONSORT, PRISMA, ARRIVE or STROBE | 추정 |
| Protocol amendments | count per study, each with a stated reason | 연구별 | Registry history and ethics committee records | 공표 통계 |
| Registered report availability | number of journals offering the format | 연간 | Center for Open Science journal list | 보고됨 |
| Results posting timeliness | months from primary completion to posted results | 연구별 | Trial registries | 공표 통계 |
| Randomisation and blinding reported | percent of eligible studies | 연간 | Risk-of-bias assessments | 추정 |
근거의 출처.
이름이 명시된 공개 데이터셋과 기관 기록으로, 각각 라이선스 등급이 있습니다. 협약에 따른 출처는 그 조건에 따라서만 사용하며 결코 재게시하지 않습니다.
| 출처 | 제공하는 것 | 라이선스 등급 |
|---|---|---|
| OSF Registries (Center for Open Science) | Time-stamped preregistrations and registered report protocols across disciplines | Open, licence per registration |
| ClinicalTrials.gov (US National Library of Medicine) | Registered trials, protocol change history and posted results | 퍼블릭 도메인 |
| WHO International Clinical Trials Registry Platform | Search across primary national trial registries | Free, WHO terms of use |
| PROSPERO (NIHR) | Registered protocols for systematic reviews | Free access, terms of use |
| protocols.io | Versioned, citable step-by-step laboratory and field methods | Open, CC BY for public protocols |
| EQUATOR Network library | Reporting guidelines by study design, with extensions | Free access, licence per guideline |
| Crossref metadata | Links between registrations, articles, corrections and funders | Open metadata, no restrictions |
누가 결정하며 근거는 무엇에 쓰이는지.
랩 prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| 결정 주체 | 질문 | 이 페이지가 제공하는 근거 |
|---|---|---|
| Research ethics committees and institutional review boards | Whether a protocol’s risks, consent and design justify approval, and whether an amendment needs fresh review | Complete protocol, sample size rationale, amendment history with reasons |
| Journal editors | Whether to accept a registered report in principle, or whether a submitted analysis matches its plan | Time-stamped registration compared line by line with the manuscript |
| 연구 재원 제공 기관 | Whether a proposal’s design can answer its question before money is committed | Power or precision rationale, control conditions, data management plan |
| Medicines and device regulators | Whether trial evidence was generated under the protocol submitted to them | Registry history, deviation logs, results posting record |
이 주제가 생산하는 산출물
- Protocol templates stating hypotheses, controls, outcomes and analysis plans before data collection
- Registration records with time stamps and version history
- Deviation logs that date and explain every amendment
- Reporting checklists completed against the relevant EQUATOR guideline
What experts ask of any protocol.
- What exactly is the hypothesis, and what result would count against it?
- What is the comparison: a control group, a baseline period, a counterfactual model or a randomised arm?
- Which outcomes are primary, which are secondary, and how will each be measured?
- What sample, duration or number of runs is needed to detect an effect of the size that matters?
- What would make the team stop early, for safety or for futility, and who decides?
- What will be published if the result is null or the experiment fails?
Why experiments drift without a protocol.
Every experiment involves choices: which observations to exclude, which covariates to adjust for, which time window to use. Each choice may be reasonable alone, but when they are made after seeing the data, the number of possible analyses grows quickly and one of them will usually show something. This is the problem methodologists call analytic flexibility, and it is a central reason why published effects often shrink or vanish when repeated.
Interdisciplinary work adds its own drivers. Disciplines differ in what they treat as a control, how they handle missing data and what level of evidence they expect. A hydrologist, an economist and an epidemiologist may each consider a design rigorous by their own standard while disagreeing about what it shows. A protocol forces those differences into the open before work starts.
Incentives matter too. Positive, novel findings are easier to publish and fund, which rewards flexibility and penalises null results. Pre-registration and registered reports are the main structural responses.
Tools protocols draw on, and where they stop.
- 01
Pre-registration
A time-stamped record of hypotheses and analysis plans. It limits post hoc changes but cannot stop a poorly chosen question.
- 02
Power and design analysis
Estimates how much data is needed to detect a meaningful effect, using 확률과 통계. It depends on assumed effect sizes that are often uncertain.
- 03
Causal designs
Randomisation, difference-in-differences, regression discontinuity and instrumental variables from 인과 추론. Each rests on assumptions that must be stated and, where possible, tested.
- 04
Acceptance criteria
Pre-agreed thresholds for when a method, model or prototype is considered to work. They make success testable but can be gamed if set too loosely.
A protocol cannot make a bad measurement good or fix a biased sample. Deviations from protocol are permitted when justified, but are reported with their reasons.
연구되는 방식과 방법이 말할 수 없는 것.
Where this page’s references lead.
이 페이지를 읽을 때 기준이 되는 기구와 문서. 상태와 판은 바뀌며, 각 링크는 발행 기구로 연결됩니다.
질문과 답변.
What is the difference between a preregistration and a registered report?
Does preregistration forbid exploratory analysis?
Which reporting guideline applies to my study?
Can a protocol be changed after registration?
Is pre-registration required for every Labs study?
Can a protocol change during a study?
Who writes the protocol?
이 페이지에서 증거를 기여하십시오.
Reports is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.
- 이메일게이트웨이 GW-GCRI이 페이지에서 보내기증거, 정정, 사례, 방법, 검토, 번역 또는 프로필 입력 자료. 이 페이지의 태그가 붙어 도착하므로, 이를 담당하는 편집자가 받게 됩니다.
- Zenodo개방형 예치기탁하고 커뮤니티에 제출하기ORCID iD로 Zenodo에 직접 기록을 게시한 뒤, 그곳의 Reports 커뮤니티에 제출하십시오. 큐레이터가 검토하며, DOI는 귀하의 것으로 남습니다.
- 보고서GCRI모든 기여 방법 보기기여 유형, 각각의 검토 방식, 받게 되는 크레디트, 라이선스, 그리고 절대 받지 않는 것.
개인정보, 민감한 인프라 세부 정보, 공유할 권리가 없는 자료는 절대 보내지 마십시오. 기여물은 사용 전에 검토됩니다. 기여물을 보낸다고 해서 저작자 지위나 편집 권한이 부여되지 않으며, 후원으로 조사 결과를 얻을 수 없습니다.
방법, 데이터세트 또는 질문을 가져오십시오.
정보와 문의는 오늘부터 열려 있습니다. 문의에는 참조 번호가 부여되고 다음 단계가 안내됩니다. 신청, 결제, 컴퓨팅 할당, 등록은 해당 팀, 약관, 심사 역량이 갖추어졌을 때만 열립니다.
법률 문서.
이 사이트의 지도, 데이터와 그림은 맥락과 일반 정보로 제시됩니다. 조언, 예보 또는 경보가 아니며, 정보의 한계는 면책 고지와 데이터 고지에 나와 있습니다.