1. Create a project snapshot
Start with a working title that is specific rather than promotional. Then name the research type, intended output, and ethics status. Your project may be a literature review, systematic review, meta-analysis, quantitative study, qualitative study, mixed-methods project, theoretical/humanities analysis, or computational project. The intended output might be a poster, abstract, manuscript draft, preprint, journal submission, or internal review product.
2. Move from topic to gap to question
Topic
Name a real scholarly conversation, not a broad interest. “Sleep variability and academic performance” is more useful than “sleep is important.”
Gap
State what is unknown, contradictory, undermeasured, under-synthesized, or poorly translated.
Question
Write an answerable question that implies a method, dataset, text corpus, population, or evidence base.
Boundary
State what the project will not answer. Narrow scope is a sign of rigor, not weakness.
3. Let the question determine the method
| If your question asks... | A possible best-fit method | Define before beginning |
|---|---|---|
| What has already been found? | Structured literature review | Databases, search strings, inclusion rules, extraction fields, synthesis themes. |
| How strong is the effect? | Meta-analysis | Effect metric, variance extraction, heterogeneity plan, sensitivity rules. |
| What predicts or correlates? | Quantitative observational analysis | Variables, missingness, covariates, model specification, visualization. |
| How do people describe or experience something? | Qualitative analysis | Sampling/corpus, codebook, reflexivity, discrepant cases, audit trail. |
| How does a text, image, or artifact make meaning? | Humanities/media analysis | Corpus boundary, theoretical lens, unit of analysis, counter-reading. |
| Can a tool, pipeline, or model be built? | Computational project | Input data, comparison baseline, validation target, reproducibility plan. |
4. Build a traceable literature search
Define three to five concept blocks and several synonyms for each. Save the exact search strings, database, date, filters, and result count. Decide inclusion and exclusion rules before screening and keep a record of useful sources that were excluded so you do not have to reconstruct scope decisions from memory. Separate background/theory sources from direct evidence sources.
5. Define what the project will actually observe
Important concepts should be defined in measurement or evidence terms rather than dictionary language. Specify the population or corpus, the outcome or evidence that would strengthen or weaken the claim, and any comparison or counterexample. This step makes the proposed analysis more concrete and helps expose vague questions before time is invested in them.
6. Handle ethics before data collection
Pause for advisor or IRB review when a project involves surveys, interviews, identifiable private data, educational records, health data, or sensitive topics. Consent language should be considered before survey questions are finalized. Existing datasets still require attention to licensing, data-use terms, de-identification, and redistribution. If AI-assisted tools are used, keep a disclosure record of the tool, task, human verification, and final responsibility.
7. Define milestones that tell you when a phase is done
| Milestone | Definition of done |
|---|---|
| Question lock | One-sentence question, one-sentence gap, and one-sentence contribution. |
| Search log complete | Databases, strings, dates, result counts, filters, and exclusion categories saved. |
| Evidence matrix complete | All retained sources coded with the same fields. |
| Methods written | A reader could reproduce the procedure without asking what happened. |
| Poster/manuscript package | Abstract, figures/tables, references, limitations, and a submission target are ready for internal review. |
Helpful standards & references
These are starting points, not substitutes for the requirements of a specific journal, conference, institution, or research team.
EQUATOR Network reporting guidelines: https://www.equator-network.org/reporting-guidelines/
ICMJE Recommendations: https://www.icmje.org/recommendations/