How to conduct a systematic literature review

A systematic literature review is one of the most rigorous and time-consuming types of academic writing. It is also one of the most valuable. A well-conducted SLR maps the entire body of evidence on a topic, identifies what is known, what is not known, and where the gaps are. It provides the foundation for your research question and, in many fields, is a publishable contribution in its own right.

Unlike a traditional (narrative) literature review, an SLR follows a structured, transparent, and reproducible protocol. Every step is documented. The search strategy, the inclusion and exclusion criteria, the screening process, and the synthesis method are all reported in enough detail that another researcher could repeat the process and arrive at similar results.

This guide walks through the seven stages of conducting an SLR, from formulating your research questions to presenting the synthesis.

Stage 1: Define Your Research Questions

The SLR begins with one or more clearly defined research questions. These questions determine your search strategy, your inclusion criteria, and the scope of your review. Vague questions produce unmanageable search results. Specific questions produce focused, useful reviews.

A common framework for structuring SLR research questions is PICO (Population, Intervention, Comparison, Outcome), used primarily in health sciences, or PICo (Population, Interest, Context), used more broadly in social sciences and education. Your questions should specify what you are looking for, in what population or context, and what outcomes or aspects you are examining.

“What are the effects of mobile learning on student achievement in higher education?” is a well-structured question. “What is mobile learning?” is too broad for an SLR.

Stage 2: Develop Inclusion and Exclusion Criteria

Before you search, define what will be included and what will be excluded. This prevents bias: you are not choosing papers based on whether they support your argument, but based on pre-set criteria applied consistently to all results.

Common criteria include publication date range (e.g., 2015 to 2025), language (e.g., English only), study type (e.g., empirical studies only, excluding opinion pieces), population (e.g., university students only), and publication type (e.g., peer-reviewed journal articles only, excluding conference abstracts).

Write these criteria down before you search. Include them in your methodology section. Reviewers will check whether your criteria are justified and whether you applied them consistently.

Stage 3: Conduct the Database Search

Search multiple databases to ensure comprehensive coverage. The specific databases depend on your discipline. Scopus and Web of Science cover most fields. PubMed is essential for biomedical research. IEEE Xplore covers engineering and computer science. ERIC covers education. PsycINFO covers psychology.

Develop a search string using Boolean operators (AND, OR, NOT) and wildcards. Document the exact search string used for each database, along with the date of the search. “mobile learn* AND (higher education OR university) AND (achievement OR performance)” is an example of a structured search string.

Run the same search string (or an adapted version) across all selected databases. Record the number of results from each database. This information will appear in your PRISMA flow diagram.

Stage 4: Screen and Select Studies

Screening typically happens in two rounds. In the first round, screen titles and abstracts against your inclusion and exclusion criteria. This removes obviously irrelevant results. In the second round, read the full text of the remaining papers and apply the criteria more carefully.

Use a reference manager (Zotero, Mendeley, or EndNote) to manage your results and remove duplicates across databases. Some researchers use dedicated SLR tools like Rayyan, Covidence, or CADIMA for screening, which allow multiple reviewers to screen independently and compare decisions.

If possible, have a second reviewer independently screen a sample (or all) of the papers. Inter-rater reliability demonstrates that your screening decisions are not subjective. Report any disagreements and how they were resolved.

Record the reason for excluding each paper during full-text screening. “Excluded: not empirical (opinion piece).” “Excluded: population is secondary school, not higher education.” These exclusion reasons are reported in the PRISMA flow diagram.

Stage 5: Extract Data

Create a data extraction form or spreadsheet before you begin reading. The form should capture the information you need to answer your research questions. Common fields include author(s), year, title, journal, country, research design, sample size, population, key findings, and quality assessment score.

Extract data consistently across all included studies. If two reviewers are involved, both should extract data independently and reconcile any discrepancies.

The data extraction spreadsheet becomes a core reference document for your review. It is often included as an appendix or supplementary material in the published review.

Stage 6: Assess Quality

Not all studies are equally rigorous. A quality assessment evaluates the methodological strength of each included study. Several standardised tools exist depending on study type: the Cochrane Risk of Bias tool for randomised controlled trials, the Newcastle-Ottawa Scale for observational studies, the CASP checklists for qualitative studies, and the MMAT (Mixed Methods Appraisal Tool) for mixed-methods studies.

Quality assessment does not necessarily mean excluding low-quality studies. It means acknowledging their limitations and weighting their findings accordingly in your synthesis. Some SLRs include all studies but conduct a sensitivity analysis to see whether removing lower-quality studies changes the overall findings.

Stage 7: Synthesise and Present the Results

The synthesis is where you bring everything together. There are two main approaches.

Quantitative synthesis (meta-analysis) combines the numerical results of multiple studies to calculate a pooled effect size. This is appropriate when the included studies are sufficiently similar in design, population, and outcome measures. Meta-analysis requires statistical expertise and is most common in health sciences and psychology.

Narrative (thematic) synthesis organises the findings by theme, pattern, or category. This is appropriate when the included studies are too heterogeneous for statistical pooling. You identify the key themes across studies, compare and contrast findings, note areas of agreement and disagreement, and discuss possible explanations for inconsistencies.

Present the screening process using a PRISMA flow diagram. This is a standardised four-stage flowchart showing the number of records identified, screened, assessed for eligibility, and included, with exclusion reasons at each stage. Most journals and thesis committees expect a PRISMA diagram for any SLR.

Present the characteristics of included studies in a summary table. Present the synthesis (whether quantitative or narrative) in the results section, organised by theme or research question.

Writing Up the SLR

A published SLR follows the PRISMA reporting guidelines, which specify what should be reported in each section of the paper. The introduction should justify why the review is needed and state the research questions. The methodology should describe the search strategy, databases, inclusion/exclusion criteria, screening process, data extraction, quality assessment, and synthesis method. The results should present the PRISMA diagram, study characteristics table, and the synthesis. The discussion should interpret the findings, compare them to previous reviews, discuss limitations, and suggest directions for future research.

Clarity is essential. The methodology section of an SLR is dense and procedural. If it is poorly written, the reviewer cannot evaluate whether your review is rigorous. For non-native English speakers, this is where professional editing has the greatest impact.

Writing up your systematic literature review? PM Proofreading’s journal editing service ensures your SLR is clear, well-structured, and publication-ready. Every order includes tracked changes and a free language editing certificate. Upload your manuscript and get a quote.

Further Reading

• How to Write an Effective Introduction Section
• How to Structure and Organize a Research Paper
• Academic Citations and References: A Complete Guide
• How to Find the Right Journal for Your Research Paper

Frequently Asked Questions

What is the difference between a systematic review and a narrative review?

A systematic review follows a documented protocol with predefined research questions, inclusion/exclusion criteria, and a transparent search strategy. It is reproducible and aims to minimise bias. A narrative review is more flexible: the author selects and discusses relevant literature without a standardised protocol. Systematic reviews are considered stronger evidence because their methodology is transparent and repeatable.

How long does a systematic literature review take?

A typical SLR takes 6 to 18 months from protocol development to final manuscript. The search and screening phase alone can take two to four months depending on the number of results. Writing up takes another two to four months. SLRs with meta-analysis require additional time for statistical analysis. Plan accordingly and start early.

Do I need to use the PRISMA guidelines?

Yes, for any systematic review submitted to a peer-reviewed journal. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) is the accepted standard for reporting SLRs. Most journals require a PRISMA flow diagram and adherence to the PRISMA checklist. For theses, check your university’s requirements, but using PRISMA is always good practice.

Can I conduct an SLR alone or do I need a team?

You can conduct an SLR alone, and many PhD candidates do. However, having a second reviewer for screening and data extraction strengthens the reliability of your review. If a second reviewer is not available, document your screening decisions thoroughly and consider having a colleague check a random sample of your decisions for consistency.