Reference

Last updated on 2026-10-03 | Edit this page

Glossary

Reproducibility: Obtaining the same results using the same data and the same analysis steps.
Replicability: Achieving similar results with new data.
Generalization: Applying results to different contexts or populations.
Verification: The process of checking that a study’s findings are accurate and free from errors.
Research Integrity: Adherence to ethical and professional principles in the conduct of research, which is often supported by practices like reproducibility.
Open Science: The practice of making research outputs (data, code, methods, publications) openly available and accessible so that others can inspect, reuse, and build on them.
Reproducibility Crisis: The documented finding across many research fields that a substantial proportion of published results cannot be reproduced or replicated by independent researchers. Also called the replication crisis; the framing itself is debated (see episode 3).
Transparency: Making the data, code, and decisions behind a result visible to others, so a reader can trace how conclusions were reached.
Rigor: The use of methods and reporting practices that make results reliable and resistant to error or bias.
Responsible Research Conduct: The set of ethical and professional standards governing how research is planned, carried out, and reported, of which reproducible practice is one component. Closely related to Research Integrity above, but framed around researcher conduct and training rather than the outcome of trustworthy results.
Open Access: Making published research outputs freely readable online without subscription or payment barriers.
Open Data: Making the data underlying a study publicly available under terms that permit reuse. Open data can support reproducibility, but public access is not required for an authorized person to rerun a documented analysis.
Version Control: A system that records successive changes to files over time, so that any earlier state can be recovered and the history of who changed what is preserved. Git is the most widely used version control system.
Scholarly Communication: The library service area covering how research is published, licensed, disseminated, and credited.
Research Workflow: The sequence of steps a researcher takes to produce an output such as a dataset, an analysis, or a publication.
Documentation: The written record of methods, data provenance, and analysis decisions that lets another person retrace a project.
Research Data Management: The institutional service area concerned with organizing, describing, storing, and sharing research data across a project lifecycle.
Codebook: A document that records, for each variable in a dataset, its name, meaning, permitted values, units, and coding of missing data.
README File: A plain-text file stored alongside a dataset or project that describes its contents, structure, provenance, and terms of use.
Container: A packaged, self-contained computing environment that bundles software and its dependencies so an analysis runs the same way on a different machine. Not the same as a lightweight dependency-management tool like renv.
Electronic Lab Notebook (ELN): A digital replacement for a paper lab notebook, used to record experimental procedures, observations, and results in a searchable and shareable form.
Markdown: A lightweight plain-text formatting syntax used to write documents that can be converted to HTML, PDF, and other formats; the basis of R Markdown, Quarto, and HackMD.
Robust: A finding that persists when reasonable, defensible changes are made to the analysis approach - a different but equally valid statistical test, for example. Distinct from reproduced (same data, same method) and replicated (new data, same method).
Active Citation: A practice in qualitative research where a claim in a paper links directly to the specific passage of source material that supports it, so a reader can check the evidence without redoing the whole analysis.
Code Capsule: A bundle of code, data, and computing environment packaged together (e.g. via Code Ocean) so someone else can run it without recreating the original setup.
Environment Management: Recording the exact software versions and dependencies a project used so it can be re-run the same way later or on another machine. Tools range from lightweight, language-specific options like renv to full containers like Docker.
Environment Capture: Recording the specific software and versions needed to rerun an analysis - the general practice that environment-management tools automate.
De-identified: Having details removed or obscured that could reveal who a research participant is. De-identifying material is a step toward sharing it responsibly, not the same as getting explicit permission to publish it.
IRB (Institutional Review Board): The committee at a research institution that reviews and approves research involving human participants, including how their data must be stored and protected.
Data Management Plan: A document, often required by funders, describing how a project’s data will be organized, stored, and shared.
Reproducibility Statement: A description of what someone would need in order to check or rerun a piece of research, including any access limits on the underlying data.
Coding (qualitative): In qualitative analysis, attaching descriptive labels to passages of text to organize and interpret them - distinct from writing program code.
Coding Scheme: The set of labels used in qualitative coding and the agreed meaning of each one.
Auditable: Describes an analysis where another person can trace the evidence and reasoning behind a conclusion, whether or not they reach the same interpretation.
SPSS Syntax: A saved set of commands that repeats an analysis in the SPSS statistical software package.
Script: Saved computer instructions that can be re-run to repeat a task, such as a data analysis step.
Dependency: Additional software a project’s code needs in order to run.
Package: A reusable, shareable bundle of software code, such as an R or Python package.
Lockfile: A record of the exact versions of a project’s dependencies, used to recreate the same computing environment later.
Provenance: A record of where a dataset came from and how it has changed over time.
Variable: A recorded characteristic in a dataset, usually represented as a column.
Permitted Values: The set of values a variable is allowed to take, as defined in a codebook.
LaTeX: A document-preparation system commonly used for producing polished, technical documents, often via an online editor such as Overleaf.
Computational Reproducibility: Obtaining consistent results by re-running the same analysis code on the same input data.
Process Transparency: In qualitative research, documenting how material was selected, coded, and interpreted clearly enough that another researcher can trace and critically assess the reasoning, even if they reach a different interpretation.