- Prepare a multi-database search log and a read-source matrix.
Start with the evidence you need
Specify whether the question concerns experiences, associations, interventions or methods. Split it into a few concepts and list synonyms, variant spellings and disciplinary terms. Translating a local-language title word for word rarely produces a sufficient search.
Teaching example: for active learning and student engagement, create a teaching-method block and an engagement block. Add population or outcome restrictions only after testing whether they remove relevant evidence. This example is not a validated search strategy.
Translate the strategy for each database
OR commonly combines synonyms and AND joins concepts, but phrase searching, truncation and field tags differ between interfaces. Consult the database’s own guide. In PubMed, review MeSH and automatic term mapping instead of assuming that an exact phrase and a mapped query behave identically.
Use more than one discovery route
Match the database to the discipline and document type. A library catalog such as HOLLIS can help discover books and other materials, but discovery does not grant access to every licensed item. Inspect references and forward citations of key studies. A broad search engine does not establish comprehensive disciplinary coverage.
Preserve the search history
Record database, interface, execution date, full query, limits and result count. Keep the original export; deduplicate with identifiers and title checks rather than an opaque deletion rule. For systematic reviews, also preserve screening decisions and reasons for full-text exclusions. A later rerun may produce different counts because database coverage changes.
Test and deliver the strategy
Check whether known relevant studies are retrieved and inspect where irrelevant results originate. A specialist librarian can help review the strategy. Deliver the search log, original records, deduplication rules and update date. Describe coverage and limitations instead of claiming an unexplained complete search.
Practical research checklist
- Define the question and eligibility rules.
- Test syntax separately in every interface.
- Preserve exports and query versions.
- Keep discovery separate from study appraisal.
Worked case and implementation decisions
The following is a fictional teaching case. Do not use its numbers or wording as actual study findings.
Break the fictional academic-support question into student population, support and engagement blocks. Combine synonyms within blocks with OR and concepts with AND. Translate syntax and fields for each database; the same pasted query can behave differently across interfaces. Test known relevant papers to detect an overly restrictive strategy and log revisions. General web search helps discovery but alone is not a complete systematic-review strategy. Preserve exports and distinguish discovery from reading and appraisal.
| Stage | Teaching example | Verification question |
|---|---|---|
| Block | student OR undergraduate | Are local terms and synonyms considered? |
| Combination | population AND support AND engagement | Are logic and parentheses correct? |
| Database | Name, interface and fields | Is syntax interface-specific? |
| Check | Known relevant study | Does the query retrieve it? |
| Log | Date, query, count and export | Is it repeatable and auditable? |
Exercise output: Prepare a multi-database search log and a read-source matrix.
Sources and further reading
Official sources for verification and further reading