RESEARCH QUESTION -> DESIGN -> ETHICS
ALL YEARS | PAGE 81A - QUESTION -> DESIGN -> GOVERNANCE -> FEASIBILITY -> PROTOCOL
A. RESEARCH-TO-SCHOLARSHIP OPERATING SYSTEM
| Stage | Core decision | Required output |
|---|---|---|
| 1. Question + stakeholders | Define the patient/population, exposure/intervention, comparator, outcome or phenomenon; include relevant patient/community perspective. | One answerable question + why it matters. |
| 2. Design match | Select experimental, observational, diagnostic, qualitative, evidence-synthesis or improvement design based on the question. | Design capable of answering the question - not merely using available data. |
| 3. Governance | Determine research vs QI/program evaluation; human-subjects status; privacy/data-use requirements; local approvals. | Written institutional determination/approval before regulated work starts. |
| 4. Protocol | Predefine variables, outcomes, confounders, eligibility, sampling, analysis, data management and roles. | Versioned protocol + analysis/data plan. |
| 5. Execute + document | Capture data consistently; log deviations, missingness, code versions and decisions. | Auditable dataset + provenance trail. |
| 6. Analyze + interpret | Estimate effects/associations with uncertainty; test assumptions; separate confirmatory from exploratory work. | Claims proportional to design and evidence. |
| 7. Disseminate | Use the reporting guideline, authorship/contributorship rules, transparent disclosures and appropriate venue. | Abstract/poster/manuscript that another team can understand and evaluate. |
B. QUESTION -> DESIGN MAP
| Question | Typical design family | Key caution |
|---|---|---|
| Does an intervention cause benefit/harm? | Randomized trial when feasible; otherwise carefully designed comparative observational study. | Feasibility does not erase confounding. |
| What predicts future outcome? | Prospective/retrospective cohort. | Separate prediction from causal inference. |
| What exposures are associated with a rare outcome? | Case-control. | Control selection can create major bias. |
| How common is a condition/exposure now? | Cross-sectional/prevalence study. | Temporal sequence is often limited. |
| How accurate is a test? | Diagnostic-accuracy study with appropriate reference standard. | Spectrum/verification bias can distort performance. |
| How do people experience a process? | Qualitative or mixed-methods study. | Sampling and analytic rigor still matter. |
| What does the total evidence show? | Systematic review +/- meta-analysis. | Protocol/search completeness and heterogeneity matter. |
| Can a local care process improve? | QI/improvement study with iterative measurement. | Obtain the appropriate institutional determination; publication intent alone does not define research. |
TOP 10 RESEARCH + QI CONCEPT - 1: THE QUESTION DECIDES THE DESIGN - NOT THE DATASET OR FAVORITE TOOL
| DEEP EXPLANATION | WHY STUDENTS MISS IT | EASIEST CLUE |
|---|---|---|
| A strong project begins by stating what decision the work should inform. The design follows the causal, | Accessible data and familiar software create momentum. Learners can confuse analytic convenience | Before opening the dataset, finish: "If I knew the answer to ___, I would understand/decide ___." Then choose |
| descriptive, predictive, qualitative or improvement question. Starting with "I have this dataset" often | with scientific fit. | the design. |
| creates a study that can calculate something but cannot answer the intended question. | ||
| TOP 10 RESEARCH + QI CONCEPT - 2: RESEARCH VS QI IS A GOVERNANCE QUESTION - NOT A PUBLICATION LABEL | ||
| DEEP EXPLANATION | WHY STUDENTS MISS IT | EASIEST CLUE |
| Many local QI activities are not human-subjects research, but some improvement projects also have a | "QI" sounds operational and low-risk, while "research" sounds formal. The label can be chosen for | Ask: "Who at my institution is authorized to determine the regulatory pathway?" Get that determination before |
| research purpose. Intent to publish by itself does not determine status. Use the institution's IRB/QI/privacy | convenience instead of applying the governing definitions and local process. | regulated work begins. |
| pathway and obtain the required determination rather than self-classifying. | ||
| TOP 10 RESEARCH + QI CONCEPT - 3: A PROTOCOL BEFORE ANALYSIS REDUCES OUTCOME-CHASING | ||
| DEEP EXPLANATION | WHY STUDENTS MISS IT | EASIEST CLUE |
| Predefining outcomes, exclusions, covariates and the main analytic plan makes later deviations visible. | Software makes it easy to try many models and remember only the compelling result. Retrospective | Write the primary question, outcome, major exclusions and analysis before viewing the answer. Version every |
| Exploratory work is still valuable, but it should be labeled as exploratory rather than silently becoming the | rationalization can feel like normal refinement. | later change. |
| original hypothesis after results are seen. |
TOP 10 RESEARCH + QI CONCEPT - 1: THE QUESTION DECIDES THE DESIGN - NOT THE DATASET OR FAVORITE TOOL
| OHRP QI FAQ | OHRP 2018 decision charts |
|---|---|
| 2026/27 research + scholarship expanded edition - verify local IRB/QI/privacy policy, sponsor requirements and current regulations before use |