Did You Use Enough Search Terms to Reduce Vocabulary Bias?

Could Relevant Studies Be Using Words You Never Searched?

You know what your research topic is called. The difficulty is that other researchers may not call it the same thing.

A concept can appear under synonyms, older terminology, acronyms, spelling variants, technical labels, colloquial expressions, brand names, or terminology specific to another discipline. Databases may also assign controlled vocabulary that differs from the language authors use in titles and abstracts.

This creates vocabulary bias: your search may systematically favor studies using the terminology you anticipated while making conceptually relevant studies expressed in other language harder to retrieve.

The solution is not to accumulate every remotely related word. The real question is whether your search represents the important ways the concepts in your research question are likely to appear in the literature.

Enough Search Terms Means Enough Ways to Express the Concept

Why One Concept Usually Needs More Than One Search Term

Authors do not share a single vocabulary

Scientific concepts are rarely described with perfect terminological consistency. Different disciplines can use different names for similar phenomena. Terminology can change over time. Abbreviations compete with full expressions. British and American spelling may differ. Researchers may also use broad, narrow, technical, or population-specific terms for substantially overlapping concepts.

A search based only on the wording familiar to you can therefore retrieve a linguistically convenient subset of the literature rather than the full conceptual territory you intended to search.

Free-text terms and controlled vocabulary solve different problems

Many bibliographic databases support both free-text searching and controlled vocabulary. Free-text searching looks for words occurring in searchable record fields such as titles and abstracts. Controlled vocabulary uses standardized subject terms assigned to records, such as Medical Subject Headings (MeSH) in MEDLINE.

These approaches complement each other. Controlled vocabulary can connect records that describe the same subject using different wording. Free-text searching can retrieve terminology not adequately represented by indexing, including emerging concepts, recent records awaiting full indexing, and expressions used by authors themselves. Cochrane guidance therefore recommends an appropriate combination of controlled vocabulary and free-text terms when constructing searches for intervention reviews.

Build terms around concepts, not around the sentence of your research question

A research question may contain many elements, but that does not mean every element belongs in the database search. Some concepts, particularly outcomes or comparators, may be inconsistently mentioned in titles, abstracts, or indexing. Requiring them can make the search unnecessarily restrictive.

Cochrane explicitly cautions that searching every aspect of a review question may be undesirable. Search construction should instead identify the main concepts that can usefully distinguish relevant records, then develop appropriate terms within those concepts.

OR usually broadens within a concept; AND narrows across concepts

Suppose one concept can be expressed as "artificial intelligence," "machine learning," or "AI." Combining those alternatives with OR allows a record containing any of them to qualify for that concept. Adding a genuinely useful synonym can therefore increase sensitivity.

AND behaves differently. It requires the connected concepts to occur together. Adding another concept with AND can exclude records that would otherwise have been retrieved. This is why "more search terms" is an ambiguous goal. More alternatives within an existing concept can broaden retrieval, while more mandatory concepts can narrow it.

Vocabulary should be discovered, not merely brainstormed

Your first list of keywords is a starting point. Useful additional terms can be found by examining known relevant papers, database thesauri, indexing assigned to relevant records, terminology used in systematic reviews on related questions, and the language appearing in records retrieved during exploratory searches.

This makes search development iterative. Cochrane's methodological standards explicitly describe search strategies as something that may be modified in response to what retrieval reveals.

Look beyond simple synonyms

Vocabulary variation can take several forms. Depending on the concept and database, you may need to consider acronyms, singular and plural forms, hyphenation, spelling variants, former terminology, scientific and common names, word endings, or phrases whose words can appear close together but not in a fixed order.

Truncation, wildcards, phrase searching, and proximity operators can sometimes capture these variations efficiently. They should still be used deliberately. Truncating too early in a word, for example, may retrieve large numbers of unrelated terms.

The same vocabulary strategy cannot simply be pasted into every database

Databases differ in controlled vocabularies, field codes, truncation symbols, proximity operators, phrase handling, and other search syntax. MeSH and Emtree, for example, are not interchangeable indexing systems. A concept represented by one subject heading in MEDLINE may require a different heading in Embase.

Search strategies therefore need translation, not merely copying. The underlying concepts should remain consistent while the syntax and controlled vocabulary are adapted to the database being searched.

Test whether the search can retrieve studies you already know matter

If you have several independently identified relevant studies, check whether your strategy retrieves them. A missed study can be diagnostically useful. Perhaps the authors used an unexpected synonym. Perhaps the record was indexed under a different subject heading. Perhaps one of your AND concepts was absent from its title and abstract.

This does not prove that a search retrieving every test paper is comprehensive. Known studies are only a small sample of the possible literature. The exercise can, however, expose vocabulary gaps before the final search is run.

Peer review can reveal terms and errors you overlooked

For consequential systematic searches, consider having the strategy reviewed by an experienced information specialist or librarian. The PRESS framework specifically examines subject headings, free-text terms, spelling variants, synonyms, truncation, Boolean logic, and other elements of electronic search strategies.

PRISMA-S also recommends describing any search peer-review process and reporting complete strategies exactly as run. This makes the vocabulary choices visible rather than reducing the methods to a statement that "relevant keywords were used."

How a Narrow Vocabulary Can Quietly Miss Relevant Studies

The improvement comes from better conceptual representation, not from reaching a target number of keywords. Ten carefully chosen alternatives may outperform thirty loosely related terms, while a complex or inconsistently described concept may legitimately require many more.

Common Mistakes When Choosing Search Terms

Build a Vocabulary Map Before You Trust Your Search

For each essential concept, ask how different researchers might express it. Then investigate those possibilities rather than relying entirely on intuition. Your search vocabulary should be broad enough to accommodate genuine linguistic variation but disciplined enough to remain connected to the concept you intend to retrieve.

Vocabulary breadth cannot compensate for searching the wrong information sources. If relevant evidence sits outside the databases you selected, even a superb set of synonyms will not retrieve it. Search-term coverage and database coverage should therefore be evaluated separately.

Before You Finalize Your Search Vocabulary

Questions About Search Terms and Vocabulary Coverage

Search the Concept, Not Just Your Favorite Word for It

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.