Literature Review and Desk Research

What is the difference between desk research and a literature review?

In practice they are the same activity applied in different settings. Desk research usually describes commercial work drawing on statistics, industry and company sources; a literature review usually describes academic work drawing on peer-reviewed publications. The method here covers both, because the appraisal problems are identical.

How do I judge whether a research source is credible?

Score it on eight dimensions separately rather than forming a general impression: authority, recency of the underlying data (not the publication date), methodology, sample and base, geography, relevance to your actual question, conflicts of interest, and whether it is primary evidence or a restatement. A fatal weakness on one dimension is not offset by strength on another.

How do I check whether an AI-generated citation is real?

Open the source. A reference is real when you can retrieve the document and confirm that the locating details and the specific claim both come from it, not from something citing it. The characteristic AI failure is a half-remembered reference: roughly the right author, roughly the right year, a plausible journal, and no such paper. Treat any reference you cannot open as absent, not as a lead.

Can AI do a literature review?

It can do the mechanical parts well: building search vocabularies across disciplines, screening at scale, extracting consistently, and holding a large evidence table in view. It cannot be trusted to supply sources from its own knowledge, and it should never generate a bibliography without access. Verification, quality appraisal and the sufficiency judgement need a researcher.

Two published sources give different numbers. Which one do I use?

Usually neither, until you have diagnosed the difference. Most conflicts turn out to be definitional (different constructs under the same word), methodological (stated importance versus observed behaviour), sampling, temporal or geographic rather than genuine contradiction. Report both figures with their sources and say what causes the gap. Do not average them.

How do I know when to stop searching?

Write the stopping rule before you start. A workable one: stop when each review question has either converging evidence from at least two independently produced sources of adequate quality, or a documented record of the searches that failed to find any.

When do I need a systematic review instead?

When the work will be examined, peer reviewed, or must be reproducible by a third party. A systematic review requires a pre-specified protocol, an exhaustive search, and protocol-standard reporting. A purposive documented review is a different and lighter method, and it should never be described as systematic.

How do I decide whether primary research is still needed?

Classify each review question as answered, partially answered, contested or absent, based on the quality and relevance of what you found. The questions that remain contested or absent, and that the decision actually turns on, are the brief for the primary study. Everything already established is money you do not need to spend.