Search Strategy for Systematic Literature Review—S3

A strong systematic literature review search strategy hinges on building a transparent, reproducible query—starting from the research question and then iteratively testing how keyword choices and database filters change the results. The process begins by extracting the core concepts from the question and turning them into a keyword set that includes synonyms, related terms, and spelling variations. For corporate social responsibility (CSR), that means pairing “corporate social responsibility” with interchangeable labels such as “social responsibility,” “business social responsibility,” “business sustainability,” and “corporate social performance,” while also accounting for variations like “behavior” vs. “behaviour” and “organizational” vs. “organisational.” These variations matter because relevant studies may use different terminology even when they address the same underlying construct.

Once the keyword universe is defined, the strategy uses Boolean operators—AND, OR, and NOT—to control breadth. OR expands coverage by capturing alternative terms, while AND narrows results to papers that combine concepts, and NOT can exclude irrelevant themes. The approach also incorporates practical limits aligned with the review scope, such as date ranges (e.g., last 5 years or since 2020), language filters (e.g., English-only), and document-type filters. Crucially, every decision needs documentation: which keywords were used, which databases were searched, what filters were applied, and how many results appeared at each stage. Tracking changes—such as how results shift when switching keywords, databases, or filters—supports transparency and reproducibility, and it can be summarized in a table showing what changed and what effect it had.

Search strings are then tailored to specific databases. In Google Scholar, the default behavior often treats AND as the default operator, so queries can be built by combining concept blocks with AND and using OR to capture term variants. The transcript illustrates this with a CSR question focused on antecedents and outcomes: the query can require CSR-related terms in the title, then add OR-linked antecedent/outcome-related mechanisms such as “impact,” “role,” “mediation,” or “moderation,” and optionally restrict the time window. For other platforms like Web of Science or Scopus, the same logic is applied but the string structure is adapted to the database’s syntax.

When access to major indexes is limited, the strategy can still be strengthened by cross-checking journal eligibility. One workaround described is using the Master Journal List (associated with Web of Science) to decide whether journals found via Google Scholar should be included; for example, a journal like “Administrative Sciences” can be included if it appears in the list, while a journal not found there can be skipped. Database selection also depends on discipline: business and management reviews may prioritize Scopus or Web of Science (and Business Source Complete), social sciences can use PsycINFO and Sociological Abstracts, and health sciences can use PubMed. Gray literature—reports, dissertations, and conference proceedings—can be avoided to reduce publication bias.

Finally, searching stops when the strategy proves effective. Effectiveness is indicated when results include major studies and recommended sources, and when quick checks of retrieved papers confirm they meet inclusion criteria. Additional “snowballing” steps help catch misses: scanning the reference lists of included papers and using citation tools like “cited by” to find newer studies that build on them. The overall goal is not just to retrieve papers, but to do it in a way that can be audited, repeated, and justified in the review’s methodology.