- Estimand
- The quantity a study aims to estimate for a defined population, condition and time; not simply a test name.
Dr. Didgar Research Institute
Research glossary
40 research terms with concise definitions and practical guide links, from CI and TOST to FAIR and Registered Reports.
40
- Confounding
- A causal structure that distorts an exposure–outcome comparison relative to the target effect; adjustment requires assumptions and timing.
- p-value
- Under a specified null model, the probability of data at least as incompatible with that model; not the probability that the null is true.
- Confidence interval (CI)
- An interval produced by a procedure with specified long-run coverage under its assumptions; not automatically a probability for the observed interval.
- Standard error (SE)
- Uncertainty of an estimator across sampling repetitions; different from individual-data variability or SD.
- Standard deviation (SD)
- A measure of value dispersion; specify sample/population definition and units.
- Smallest effect size of interest (SESOI)
- A meaningful-effect threshold justified scientifically or practically with explicit units before results.
- Two one-sided tests (TOST)
- An equivalence procedure against two bounds; failure to reject a difference test does not substitute for it.
- Statistical power
- Probability of rejecting a null under a specified effect and design scenario; inseparable from its assumptions.
- Missing completely at random (MCAR)
- An assumption that missingness is independent of observed and unobserved values; not implied by a small missing percentage.
- Missing at random (MAR)
- Missingness does not depend on missing values after conditioning on suitable observed information.
- Missing not at random (MNAR)
- Missingness remains related to unobserved values beyond an MAR assumption; sensitivity matters.
- Preregistration
- A time-stamped record of plans and decisions, with prior-data access disclosed and changes/exploration distinguished.
- Registered Report (RR)
- A publication format reviewing the design before results with conditional later stages; registration alone is not that review.
- In-principle acceptance (IPA)
- A Stage 1 Registered Report decision with conduct/reporting conditions; not unconditional acceptance of any later work.
- Multiverse analysis
- Transparent examination of defensible decision-dependent paths; counting significant paths is not a scientific vote.
- Computational reproducibility
- Recreating output from specified data, code and environment; distinct from a new-data replication.
- Replication
- A new-data investigation with suitable design to examine robustness or scope.
- FAIR principles
- Findable, accessible, interoperable and reusable; not necessarily unrestricted data openness.
- Data dictionary
- Variable definitions, units, types, ranges, missingness and provenance needed to interpret data.
- Data management plan (DMP)
- A plan for production, protection, documentation, access and preservation with actual responsibilities and restrictions.
- Digital Object Identifier (DOI)
- A research-object identifier; syntax alone establishes neither existence, correctness nor scientific quality.
- ORCID researcher identifier
- A researcher identifier reducing name ambiguity; not an assessment of work quality.
- Preprint
- A publicly available manuscript before journal publication with its actual review status stated.
- Accepted Author Manuscript (AAM)
- The accepted manuscript incorporating review revisions before publisher typesetting; deposit rights need checking.
- Version of Record (VoR)
- The formal publisher-record version, distinct from an AAM or preprint.
- Data Descriptor
- An article type focused on provenance, quality and data reuse under outlet-specific requirements.
- Constraints on Generality (COG)
- A statement of populations, settings, stimuli and times covered by evidence and conditions for expected extension.
- PRISMA review reporting
- A family of review-reporting guidance; a reporting format alone does not validate review conduct.
- Living systematic review (LSR)
- A review with planned ongoing surveillance, update decisions, versions and a stopping policy.
- CONSORT trial reporting
- Randomized-trial reporting guidance; check the 2025 version and design-appropriate extensions.
- SPIRIT trial protocols
- Guidance on trial-protocol content, distinct from reporting actual results.
- CHART health-chatbot reporting
- Guidance for evaluating generative-chatbot health advice and responses within a specific scope.
- COREQ qualitative reporting
- Reporting guidance for interviews and focus groups; applicability depends on the actual method.
- TOP 2025 transparency
- A transparency and verification framework; keep version and evidence levels distinct from earlier criteria.
- Researcher reflexivity
- Documented attention to researcher position, assumptions and decisions in producing and interpreting data.
- Sensitivity analysis
- Examination of conclusions under defensible alternative assumptions or decisions, not selection of a preferred result.
- Risk of bias
- Potential distortion from design or conduct, distinct from journal reputation or anonymous review.
- Similarity and attribution
- Tool matches require text interpretation; an aggregate percentage is not an automatic misconduct decision.
- Quarto computational document
- A document connecting code and narrative; dependencies and output verification still matter.