A research gap is not a subject nobody has written about. Anything nobody has written about is usually something nobody needs. A gap is a question the studies that already exist fail to answer — and the useful move is to say why they fail to answer it.
That “why” is the part most advice skips, and it is the part that turns a vague sense of novelty into a question you can actually put in a proposal.
Treat the future-work section as a starting point
The standard advice is to read the “future work” or “implications for research” section of a review and take your question from there. The evidence says that section is usually too vague to act on.
Clarke and colleagues audited 2,535 Cochrane reviews [2], which are unusual in having a dedicated “Implications for Research” section. Twelve percent of those reviews failed to specify any of the three basic components of a research question — population, intervention, outcome. Only about 17% described a gap in terms of all three.
Robinson and colleagues found the same thing when they looked at twelve evidence reports from US Evidence-based Practice Centers [5]: none of the reports described how their research gaps had been identified at all.
So the future-work section is the right place to look. It is simply not a place where the work has already been done for you. What follows is the structure that fills the gap in the gaps.
The two things you have to say about a gap
Robinson, Saldanha and Mckoy [5] proposed a framework with exactly two elements: characterise the gap, then give the reason for it. Both halves are needed. A characterisation without a reason is a topic; a reason without a characterisation is a complaint.
Characterising is done with PICOS. PICOS stands for population, intervention, comparison, outcome and setting — the framework clinical researchers use to assemble a question.
1. Characterise it with PICOS
Describe the gap along those five axes and say which of them the existing evidence handles badly. “Nobody has studied this in adults over 75” is a population gap. “Every trial compares against placebo, never against the drug people actually take” is a comparison gap.
These five are axes for describing a gap, not kinds of gap. This matters, because a widely-repeated secondary account presents PICOS as “the five types of research gap Robinson identified”. It is a misreading, and it drops the half of the framework that does the real work. The kinds are the four below.
2. Name the reason, from four
| Reason | What triggers it |
|---|---|
| Insufficient or imprecise information | No studies, too few studies, or samples too small. Also a confidence interval wide enough to include both superiority and inferiority |
| Biased information | Risk of bias in the individual studies, an inappropriate study design, or body-of-evidence problems such as publication bias and selective outcome reporting |
| Inconsistency or unknown consistency | Effects disagree in direction or size. A single study counts here too — with one study, consistency is not unknown-and-fine, it is simply unknown |
| Not the right information | Results do not apply to the population or setting you care about, outcomes are surrogate rather than the ones that matter, or follow-up is too short |
The third row is the one that changes behaviour most often. “There is one large study, so this is settled” is not a conclusion the framework allows you to reach.
The framework was not only proposed. Robinson’s team piloted it, and a later AHRQ report evaluated it against a sample of fifty systematic reviews of randomised trials [6]. Applying it is not free: the original work reports an average of 3.5 hours per evidence report, producing a mean of 14.75 research gaps per report [5].
Trace the citations to check the gap is still open
Naming a gap is worth nothing if somebody closed it between the review going to press and you reading it. Checking that is a search problem, and the search method with the best evidence behind it is snowballing. Library guides teach it as the first move in a literature search: Kyoto University’s, for one, opens its literature-search basics with “work outward from the reference lists of what you already have” [3].
Wohlin defines it in one line [7]: snowballing means “using the reference list of a paper or the citations to the paper to identify additional papers”. Reference lists are backward snowballing; papers that have since cited it are forward snowballing. His conclusion is stronger than most people expect — snowballing as a first search strategy may be a good alternative to database searching, not merely a supplement to it.
For gap-checking, forward is the half that matters. Start at the review that suggested the gap, list everything that has cited it since, and read the ones that touch your PICOS description. If nothing does, you have evidence rather than a hunch.
When your field is not clinical
The four reasons come from medicine, and the authors say plainly that it is not clear how the framework performs on questions that are not about interventions. For qualitative and management literature, Müller-Bloch and Kranz derived six categories from a grounded-theory analysis of forty literature reviews [4]: contradictory evidence, knowledge void, action-knowledge conflict, methodological, evaluation void, and theory application void.
A seven-item list also circulates, popularised through a doctoral-workshop handout and published in book form by Adu and Miles in 2023 [1]. It is a synthesis of the two lists above rather than independent evidence. Treat the count as an authorial choice — four, six and seven are three people’s groupings of the same terrain, not three findings.
Doing the tracing in one place
The gap-checking loop is repetitive: find the review, find everything citing it, open the ones that overlap your description, write down what each one closes. It gets slow when the papers sit across four different databases.
Litlas is built for that loop. One query runs across 16 scholarly sources (OpenAlex, Crossref, DataCite, OpenCitations, ROR/ORCID, arXiv, DBLP, OpenReview, ACL Anthology, PubMed, PMC Open Access, DOAJ, CORE, OpenAIRE Graph, Unpaywall and Common Crawl), and results filter by venue, author or year. Open the citation graph on any paper and you get both directions of snowballing on one screen: what it cited, and who has cited it since. Papers you keep go into a board for the project with your own notes and tags, so the sentence “this one closes the outcome gap, this one does not” stays attached to the paper, and the board exports as BibTeX, RIS or CSL-JSON when you write the proposal up.
The free plan needs no card and works right away.
For appraising the studies you find and for the screening log itself, pair it with a tool built for that stage.
Three things to remember
- A gap is a question the existing evidence fails to answer, plus the reason it fails. Without the reason you have a topic.
- PICOS describes where the evidence falls short. The four reasons describe why. Anyone who calls PICOS “the five types of gap” has merged the two and lost the useful half.
- One study is not consistency. It is unknown consistency, which is itself a gap.
Bibliography
- Adu, Philip, and D. Anthony Miles. “Understanding the Seven Types of Research Gaps.” Chapter 5 in Dissertation Research Methods, 74–84. Routledge, 2023. https://doi.org/10.4324/9781003268154-5
- Clarke, Lorcan, Mike Clarke, and Thomas Clarke. “How Useful Are Cochrane Reviews in Identifying Research Needs?” Journal of Health Services Research & Policy 12, no. 2 (2007): 101–103. https://doi.org/10.1258/135581907780279648
- Kyoto University Library Network. “Basics of Collecting Literature” [in Japanese]. https://www.kulib.kyoto-u.ac.jp/refguide/13202. Accessed 30 August 2026.
- Müller-Bloch, Christoph, and Johann Kranz. “A Framework for Rigorously Identifying Research Gaps in Qualitative Literature Reviews.” ICIS 2015 Proceedings, paper 2. https://aisel.aisnet.org/icis2015/proceedings/ResearchMethods/2/. Accessed 22 August 2026.
- Robinson, Karen A., Ian J. Saldanha, and Naomi A. Mckoy. “Development of a Framework to Identify Research Gaps from Systematic Reviews.” Journal of Clinical Epidemiology 64, no. 12 (2011): 1325–1330. https://doi.org/10.1016/j.jclinepi.2011.06.009
- Robinson, Karen A., Ian J. Saldanha, and Naomi A. Mckoy. Frameworks for Determining Research Gaps During Systematic Reviews. Methods Future Research Needs Reports No. 2. Rockville, MD: Agency for Healthcare Research and Quality, 2011. https://www.ncbi.nlm.nih.gov/books/NBK62478/. Accessed 22 August 2026.
- Wohlin, Claes. “Guidelines for Snowballing in Systematic Literature Studies and a Replication in Software Engineering.” In Proceedings of the 18th International Conference on Evaluation and Assessment in Software Engineering, article 38. ACM, 2014. https://doi.org/10.1145/2601248.2601268