Boolean Search Operators for Research: AND, OR, NOT

What is Boolean search?

Boolean search is a way of combining keywords with operators, chiefly AND, OR and NOT, to tell a database exactly which records you want. The name comes from George Boole, the nineteenth-century mathematician whose algebra of logic treats statements as true or false. A database applies that logic to every record: does it contain this term, and that term, but not the other one?

Library catalogs and academic databases were built around this kind of searching, which is why a carefully constructed Boolean string usually works far better in them than a typed question. The same logic sits behind advanced search forms, where boxes labeled all of these words or any of these words are AND and OR in disguise. Learn the three operators once and you can search almost any research database with precision.

AND, OR and NOT: what each Boolean operator does

Each operator changes the size of your result set in a predictable way. Picture two overlapping circles, one for each term: AND gives you only the overlap, OR gives you everything in both circles, and NOT gives you one circle with the overlap cut away.

Many databases expect operators in capital letters, so type AND, OR and NOT in capitals as a habit. Typing several words with no operator between them is often treated as AND, but do not assume it: some systems treat a string of words as a phrase, or rank results by loose relevance instead.

Parentheses, phrases, truncation and proximity

The three operators become far more powerful with four companions. Support and exact symbols differ between databases, so check the help page of each one you use.

How to build a Boolean search string, step by step

The most reliable method is to work from your research question to a concept table, and from the table to a search string. Suppose your question is how social media use relates to sleep in teenagers. If you do not yet have a focused question, start with the guide to writing a good research question.

Followed through, the steps below turn that question into a string like this: (teen* OR adolescen* OR "young people") AND (sleep OR insomnia OR bedtime) AND ("social media" OR smartphone* OR "screen time"). Each pair of parentheses holds one concept, and every record returned must match at least one term from each group. Here is the method:

1. 1Pick out the two to four core concepts in your question: here, teenagers, sleep and social media.
2. 2For each concept, list synonyms, related terms, spelling variants and the technical terms researchers use.
3. 3Join the terms within each concept with OR, and wrap each group in parentheses.
4. 4Join the concept groups to each other with AND.
5. 5Add quotation marks to fixed phrases and truncation to word stems.
6. 6Run the search, read the first twenty or so titles, and adjust any term that pulls in noise or misses obvious matches.
7. 7Save the final string, the database and the date in a search log, so you can rerun or report it.

Too many results or too few? How to adjust

Your first search is a draft. Look at the result count and, more importantly, the first page of titles. The records that are exactly what you wanted teach you the most: their titles, abstracts and subject headings show the vocabulary researchers actually use, which is often different from yours. Add those terms to your concept table, because a second round built on borrowed vocabulary usually beats the first.

Beyond borrowing vocabulary, adjust according to what the first page tells you:

• Too many, mostly irrelevant: add a concept group with AND, limit key terms to the title or abstract field, or apply the database's filters for date, language or document type.
• Too many, mostly relevant: narrow your question rather than your search, then add the new concept.
• Too few: remove the least essential concept, add synonyms with OR, loosen phrases that authors may word differently, or check your truncation.
• None at all: look for a typo, an unmatched parenthesis or a stray quotation mark before deciding the topic is empty.

Boolean search in library databases, Google and Google Scholar

Library databases support Boolean logic most fully. Many add field searching, so you can require a term in the title, abstract or author field, and controlled subject headings, standardized terms assigned to each record; PubMed's MeSH headings are a well-known example. Searching a subject heading alongside your keywords catches records that describe your concept in words you did not think of.

Google and Google Scholar handle Boolean logic more loosely. Both treat a space between words roughly as AND, accept OR in capitals, use a minus sign in place of NOT, and support quotation marks for phrases. Neither handles nesting and truncation the way databases do, so keep strings for them short. The guide to using Google Scholar effectively covers its operators in detail.

Common Boolean search mistakes

Once your searches are working, the guide to finding credible sources covers how to judge what they return. Until then, these errors account for most frustrating searches, so check for them before you decide your topic has nothing written on it:

• Using AND between synonyms, as in teenagers AND adolescents, which demands both words and shrinks results.
• Overusing NOT, which removes records that discuss your topic and merely mention the excluded term.
• Mixing AND and OR without parentheses, leaving the database to guess the grouping.
• Truncating too early, so the stem matches unrelated words.
• Typing operators in lowercase in a database that reads them as ordinary words.
• Searching full sentences in a database built for keywords.
• Never checking the subject headings of good results for better terms.