How to find a research gap: 5 methods, an example and an AI shortcut

This guide is for students and early-career researchers choosing a thesis, dissertation or grant topic. It covers finding the gap. Turning the gap into a question you can answer is in the pillar guide, how to write a research question.

What is a research gap?

A research gap is a question in a field that existing studies have not answered, have answered poorly, or have answered in conflicting ways. It is the reason your study is needed: the sentence in your introduction that starts "However, no study has...".

A gap is not just "nobody has studied X". Some questions are unstudied because they don't matter, or because the answer is obvious. A useful gap is one where an answer would change what people know or do.

What are the types of research gaps?

Health researchers have a formal version. Robinson and colleagues' framework for identifying gaps from systematic reviews classifies the reason a gap exists as insufficient or imprecise information, biased information, inconsistency or unknown consistency, or not the right information [1]. Those four reasons work in any field: ask which one applies to your candidate gap.

Is gap-spotting the only way to justify a study?

No. Sandberg and Alvesson, reviewing 52 articles in organization studies, found gap-spotting was the most common way researchers built questions, and argued that it tends to extend existing theory rather than challenge it [2]. Their alternative, problematization, questions an assumption the literature takes for granted. Most theses use a gap; a strong one sometimes also challenges an assumption.

How do you find a research gap? 5 methods

Use at least two of these. Each one finds different kinds of gaps.

Method 1: Read the "future research" sections

Most papers end with limitations and suggestions for future research. Reviews do this systematically: the Cochrane Handbook asks every Cochrane review to include "Implications for research", framed by the population, intervention, comparison and outcomes still to be studied [3]. A BMJ guide proposes the EPICOT format for these recommendations: Evidence, Population, Intervention, Comparison, Outcome and Time stamp [4].

1. Find the three most recent reviews on your topic.
2. Copy every future-research suggestion into a table with its source and year.
3. Mark suggestions that appear in more than one review: those are persistent gaps.
4. Check whether anyone acted on the older suggestions (Method 3).

Method 2: Look for conflicting findings

Where studies disagree, there is usually a gap in why. Put the studies on each side in a table with their population, design, sample size, measure and result. Look for the variable that differs: age group, dose, follow-up length, measurement tool. "Studies disagree, and the difference may be X" is a gap you can test. For how to weigh studies against each other, see levels of evidence and study types.

Method 3: Follow citation trails

Take a key paper that ended with "further research is needed on X" and look at the newer papers that cite it. If none of them studied X, the gap is probably still open. If several did, the gap has closed, or moved. Our guide on how to find who cited a paper shows how in each tool, and citation snowballing extends it to a whole set.

Method 4: Map the review articles

A review summarizes what is known; its tables show what isn't. Read the characteristics-of-studies table and count: which countries, age groups, settings and designs appear, and which don't. An evidence gap map is a visual table of interventions against outcomes that shows where studies exist and where they don't. If a review in your field has one, the empty cells are candidate gaps.

Method 5: Use an AI gap finder, then verify

An AI research gap finder reads a sample of recent and highly cited papers on a topic and suggests what they leave open: unanswered questions, under-studied groups, untried methods and disagreements. It is fast and useful for a first list. It sees only the papers it retrieved, so a gap it reports may already be filled by a study outside its sample. Every AI-suggested gap needs the checks below.

Research gap example: from topic to gap

Example (invented studies and numbers, for illustration). Topic: mobile money and women's entrepreneurship in East Africa.

1. Review (Method 1). A 2023 review of 24 studies concludes that mobile money access is associated with higher business income for women, and calls for "longitudinal studies and rural samples".
2. Conflict (Method 2). Two studies find no effect for women who share a phone with a spouse; three that did not record phone ownership find an effect.
3. Citation trail (Method 3). 41 papers cite the review. Four are longitudinal, all in urban areas. None separates shared from owned phones.
4. Review map (Method 4). The review's table: 19 of 24 studies urban, 20 of 24 cross-sectional.
5. AI finder (Method 5). It suggests the same rural gap and adds "effects on business survival, not just income".

The gap: no longitudinal study has tested whether owning (not sharing) a phone changes the effect of mobile money on rural women's business survival. It is a population gap, a method gap and an explanation of conflicting evidence at once, which makes it strong.

How do you check that a research gap is real?

Search for the gap itself, as if you hoped to find the study that fills it. Most false gaps die here.

1. Write the gap as a question and search it directly in two databases, with synonyms, limited to the last three years.
2. Search registries for work in progress: ClinicalTrials.gov for trials [5] and PROSPERO for registered systematic reviews [6] in health topics; preprint servers in fast-moving fields.
3. Check theses. Your library's dissertation database often holds unpublished work on niche topics.
4. Read the newest review's search date. Studies published after it are not in it.
5. Ask your supervisor or a subject librarian whether they know of work on it.
6. Write down what you searched and when, so you can defend the gap in your proposal and at your defense.

If a study already fills the gap, you haven't wasted time: its limitations section is often your next gap.

How do you turn a research gap into a research question?

State the gap, then ask the question that fills it. Two templates:

• Gap sentence: "Although [what is known] (citations), no study has [missing element] in [population or setting], so [why it matters]."
• Question: "In [population], does [exposure or intervention] affect [outcome] compared with [comparison] over [time]?"

From the example: "Although mobile money access is associated with higher income for women entrepreneurs in East Africa, no longitudinal study has examined rural women or separated phone ownership from sharing." Question: "Among rural women entrepreneurs in Kenya, does owning a mobile phone change the association between mobile money use and business survival over two years?"

The pillar guide, how to write a research question, shows how to test that question with FINER and frame it with PICO, PCC or SPIDER. Then turn it into predictions with how to write a hypothesis.

How to find a research gap in AI for Research

AI for Research has two gap tools and three free ways to check them.

• Research gap finder. Enter a topic; it reads about ten recent and highly cited papers and suggests possible gaps, each citing the papers behind it. Signed out, it is free for 3 runs a day; signed in, it costs 3 credits.
• Thesis Suite, Research gaps (3 credits). For a thesis: it summarizes what is well studied, then lists 4 to 6 possible gaps, each with its type, why it may be a gap (cited), a confidence level with a reason, and a study that could address it. It ends with how to check them. The Research question module (3 credits) then turns a gap into candidate questions.
• Landscape (0 credits). For a topic: works per year, top topics, journals, institutions and the most-cited works, so you can see where research is thin. Compare two topics side by side.
• Citation Map (0 credits). Follow a key paper's citing papers to see whether its future-research suggestions were taken up (Method 3).
• Find Papers (0 credits). Search for the gap directly with filters for year and study type.

Limits. Both gap tools work from a sample of papers, mostly abstracts. A suggested gap is a lead, not proof that no one has studied the question. Run the checks above, and talk to your supervisor, before you commit. The free plan has 45 credits a month; see pricing.

FAQ

What is a research gap example?

Example: many trials show that a school-based sleep program improves sleep in urban secondary schools in high-income countries, but none has tested it in rural schools in low-income countries, where start times and home conditions differ. That is a population and setting gap. A study could test the same program in rural schools and measure the same outcomes, so the results can be compared.

How do I find a gap in the literature?

Read the most recent review on your topic and its 'implications for research' or 'future research' section, then compare studies that disagree, follow the citations of a key paper forward to see what is still open, and note which populations, settings and methods are missing. An AI gap finder can suggest leads. Check each one with a targeted search before you build a study on it.

What are the types of research gaps?

Common types are: an unanswered question (no study has asked it), a population or setting gap (studied in some groups but not others), a method gap (only weak designs or one method used), conflicting evidence (studies disagree), a theory gap (a theory not tested or applied here), and outdated evidence (the context has changed since the studies were done).

Can AI find a research gap for me?

AI can suggest possible gaps quickly by reading a sample of papers and noting what they don't cover or where they disagree. It can't confirm a gap is real, because it sees only part of the literature and can miss a study that already answers the question. Use AI suggestions as leads, then search for each gap directly and discuss it with your supervisor.

How do I know a research gap is worth studying?

Check three things. The gap matters: someone (patients, teachers, a field's theory) would act differently if it were filled. It is real: a targeted search, including trial and review registries, finds no study answering it. And you can fill it: you can reach the participants or data, with the methods and time you have. If any check fails, narrow or change the gap.

Next steps

• Turn the gap into a question: how to write a research question.
• Choose the theory behind it: theoretical vs conceptual framework.
• Write it up for approval: how to write a research proposal.
• Plan the study that fills it: how to write a methodology chapter.

Sources

1. Robinson, K. A., Saldanha, I. J., & McKoy, N. A. (2011). Development of a framework to identify research gaps from systematic reviews. Journal of Clinical Epidemiology, 64(12), 1325–1330. https://doi.org/10.1016/j.jclinepi.2011.06.009
2. Sandberg, J., & Alvesson, M. (2011). Ways of constructing research questions: gap-spotting or problematization? Organization, 18(1), 23–44. https://doi.org/10.1177/1350508410372151
3. Schünemann, H. J., Vist, G. E., Higgins, J. P. T., et al. Chapter 15: Interpreting results and drawing conclusions, section 15.6.3 Implications for research. In Cochrane Handbook for Systematic Reviews of Interventions (current version). Cochrane. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-15
4. Brown, P., Brunnhuber, K., Chalkidou, K., et al. (2006). How to formulate research recommendations. BMJ, 333(7572), 804–806. https://doi.org/10.1136/bmj.38987.492014.94
5. U.S. National Library of Medicine. ClinicalTrials.gov. https://clinicaltrials.gov/
6. Centre for Reviews and Dissemination, University of York. PROSPERO: International prospective register of systematic reviews (funded by NIHR). https://www.crd.york.ac.uk/prospero/