An evidence gap map is a structured display of where studies exist across a field. Rows might represent interventions, exposures, or programs; columns might represent outcomes or populations. The cells show the amount and sometimes the type or confidence of available evidence. A map helps scope future synthesis and research, but it does not by itself calculate whether an intervention works.
State the map’s decision purpose
Begin with the audience and action. A funder may need to spot under-studied outcomes, a policy team may need an overview before commissioning reviews, and a research group may need to avoid duplicating completed work. Write the geographic, population, time, and study-design boundaries. Without a purpose, taxonomy expands until screening becomes unmanageable.
Draft a framework before searching
Define the row and column categories using domain language and stakeholder input. Test whether categories are mutually understandable even if they cannot be perfectly exclusive. Keep a codebook with definitions, examples, exclusions, and tie-breaking rules. Pilot several known studies to reveal categories that are too broad, too narrow, or dependent on information abstracts rarely report.
Design a reproducible search
List databases, websites, date limits, languages, grey-literature sources, and exact search strings. Save export dates and deduplication rules. A map focused only on journal databases may systematically miss government evaluations, dissertations, or reports. If resources force restrictions, document the likely direction of the resulting bias.
Screen with explicit eligibility criteria
Use two stages: title and abstract, then full text. Record reasons for full-text exclusion in stable categories. When possible, use independent checking or calibration between reviewers. A machine or assistant can prioritize records, but humans remain responsible for the inclusion boundary and for detecting vocabulary the search did not anticipate.
Code evidence, not conclusions
Capture study design, setting, population, intervention, comparator, outcomes, follow-up, and review status. Decide whether one study can populate multiple cells and how clusters of related reports are linked. Do not mark a cell positive or negative merely because the authors use favorable language. Effect direction and quality need a separate synthesis process.
Add confidence without creating false precision
Some maps distinguish systematic reviews from primary studies or display an appraisal category. If you include quality information, name the tool, unit of assessment, reviewer process, and meaning of each symbol. Avoid combining incompatible quality scores into a number that looks objective. Make uncertainty legible through notes and filters.
Interpret blank and crowded cells carefully
An empty cell may reflect a true research gap, poor indexing, terminology mismatch, restricted languages, unpublished work, or an impossible category combination. A crowded cell can still lack relevant populations, long follow-up, harms, or reliable designs. Rank potential gaps by decision importance, plausibility, equity, feasibility, and what additional evidence would change.
Publish the map with an audit trail
Provide the protocol, search date, flow counts, coding framework, included-study table, appraisal method, and update policy. Give each cell a path back to its records. Version the underlying data so later additions do not erase what decision makers saw at a particular time. Accessibility matters: supply a downloadable table alongside an interactive visualization.
Frequently Asked Questions
Is an evidence gap map the same as a systematic review?
No. A map describes the distribution of evidence across a framework. A systematic review usually answers a narrower question through detailed synthesis and may be represented inside the map.
Does an empty cell prove there is no evidence?
It proves only that the applied methods found no eligible records for that cell. Search coverage, terminology, publication bias, and coding choices must be considered.
Can AI screen all records automatically?
Automation can assist prioritization and tagging, but transparent validation and accountable human decisions remain necessary, especially near eligibility boundaries.
Put the Reading Into Practice
Take a modest topic and create a three-by-four draft framework. Code five known studies, revise one ambiguous category, and write two sentences explaining why the sparsest cell may or may not be a priority. Save the codebook change as part of the audit trail.
Related Readever Pages
- Use an AI literature review workflow
- Organize research papers and notes
- Build an evidence table
- Annotate a research paper PDF
