Literature Review
Overview
Systematic procedure for conducting comprehensive literature reviews on any topic
Steps
Step 1: Define scope and criteria
Establish clear boundaries for the review:
- State the research question precisely
- Define inclusion criteria (what makes a source relevant)
- Define exclusion criteria (what disqualifies a source)
- Set time boundaries (publication date range)
- Set geographic/language boundaries
- Define source types to include (journals, grey literature, etc.)
Step 2: Develop search strategy
Create comprehensive search approach:
- Identify relevant databases for the topic
- Academic: Google Scholar, Semantic Scholar, PubMed, SSRN, arXiv
- Grey literature: Think tanks, government reports, NGOs
- Develop keyword combinations
- Core concept terms
- Synonyms and related terms
- Technical/jargon terms
- Boolean operators (AND, OR, NOT)
- Plan snowballing strategy
- Forward: papers citing key works
- Backward: references in key works
- Identify key authors and journals to monitor
Step 3: Execute initial search
Run searches systematically:
- Execute search strings in each database
- Record number of results per database
- Export citations to reference manager
- Remove duplicate entries
- Document search date and exact strings used
Target: 100-500 initial results is typical If too few: broaden search terms If too many: add specificity or date limits
Step 4: Screen by title and abstract
Apply inclusion/exclusion criteria:
- Review each title - exclude clearly irrelevant
- Review abstracts of remaining - apply criteria
- Mark uncertain cases for full-text review
- Document reasons for exclusion
Typical reduction: 70-90% excluded at this stage If excluding too few: criteria may be too broad If excluding too many: criteria may be too narrow
Step 5: Full-text review and extraction
Deep review of remaining sources: For each source:
- Obtain and read full text
- Confirm meets inclusion criteria
- Assess quality using evidence hierarchy
- Level 1: Systematic reviews/meta-analyses (strongest)
- Level 2: RCTs
- Level 3: Quasi-experimental
- Level 4: Observational studies
- Level 5: Cross-sectional
- Level 6: Case reports/expert opinion (weakest)
- Extract key information:
- Citation details
- Methodology summary
- Key findings
- Sample/population
- Limitations
- Relevance score (1-5)
- Note quotes and insights for synthesis
Step 6: Snowball and iterate
Expand search through citation networks:
- Identify highest-quality and most-cited sources
- Forward snowball: Find papers citing these
- Backward snowball: Review their reference lists
- Screen new finds through Steps 3-5
- Repeat until reaching saturation (few new relevant finds)
Also check:
- Other work by key authors
- Recent issues of key journals
- Conference proceedings in the field
Step 7: Synthesize findings
Integrate findings across sources:
- Choose synthesis method based on literature:
- Thematic: Organize by emerging themes (diverse literature)
- Chronological: Show evolution over time (historical topics)
- Methodological: Compare by research method (methods matter)
- Theoretical: Organize by theoretical framework (competing theories)
- Code sources by theme/category
- Identify patterns within and across categories
- Note areas of consensus and contradiction
- Assess overall evidence strength
- Document gaps and limitations
When to Use
- Starting a research project requiring background knowledge
- Evaluating the evidence base for a claim or intervention
- Writing background/introduction section for papers or proposals
- Understanding current state of knowledge on any topic
- Identifying gaps in research for new contributions
- Making evidence-based policy or business decisions
- Preparing for expert consultation or interviews
- Building foundation for systematic review or meta-analysis
Verification
- Search strategy used multiple databases and snowballing
- Inclusion/exclusion criteria consistently applied
- Each source assessed for quality using evidence hierarchy
- Synthesis identifies patterns, not just lists sources
- Contradictions in literature explicitly addressed
- Gaps in literature documented with implications
- Methods documented sufficiently for reproduction
Input: $ARGUMENTS
Apply this procedure to the input provided.