A literature review proceeds in six steps: clarify the research question →
develop a search strategy → collect and screen papers → assess quality →
synthesize → write the review. The most important principle is that a
literature review must be a thematic, integrated narrative — not a series of
individual paper summaries.

Why a Literature Review Is Necessary

Failing to survey prior research thoroughly leads to three critical problems. You risk repeating work that has already been done, testing hypotheses that have already been refuted or addressing limitations that have already been overcome, and being unable to demonstrate what is new about your study — a fatal weakness in thesis defenses and journal submissions.

Types of Literature Reviews

The right type depends on your research purpose.

Type	Purpose	Best For
Narrative review	General overview of a topic	Thesis introductions, broad surveys
Systematic review	Comprehensive, reproducible review	Medicine, social sciences
Meta-analysis	Statistical synthesis of multiple studies	Combining quantitative research
Scoping review	Mapping research scope and evidence	Emerging fields, planning stages

Six-Step Process

Step 1: Clarify the Research Question

A clear research question is the starting point of every literature review. If the question is vague, your search will be either too broad or too narrow, causing you to miss key sources. Use the PICO framework (Population, Intervention, Comparison, Outcome) to structure the question.

Example: "Among university students (P), does flipped learning (I), compared to traditional lectures (C), affect critical thinking skills (O)?"

Step 2: Develop and Execute a Search Strategy

Keyword selection — Express core concepts in multiple terms. For "online learning," include synonyms like e-learning, distance education, and remote learning.

Boolean operators — Combine AND (all terms), OR (any term), NOT (exclude), "" (exact phrase), and * (wildcard).

Database selection — Use at least three databases.

Database	Strength
Google Scholar	Broad coverage, free
PubMed	Biomedical focus
Scopus	Strong citation analytics
Web of Science	High-quality journal indexing
RISS	Domestic theses and journals

Inclusion and exclusion criteria — Define criteria for publication year, language, study type, and population in advance and document them. Reproducibility is critical in systematic reviews, so always record the search date, query, and result count.

NubintAI's AI Paper Search lets you type a research question in natural language and searches a database of 280 million scholarly works using semantic matching — no exact keywords required.

Step 3: Collect and Screen Papers

Follow the PRISMA flow for four-stage screening.

Identification — Collect papers through database searches
Duplicate removal — Remove papers found in multiple databases
Screening — First-pass filter by title and abstract
Eligibility assessment — Final selection through full-text review

The Literature Review Agent in deep research mode can search and analyze up to 40 papers simultaneously, organizing key findings, methods, and conclusions in a structured format. This dramatically shortens the screening stage.

Step 4: Assess Quality

Evaluate collected papers against these criteria:

☐ Is the research objective clearly stated?
☐ Is the research design appropriate for the question?
☐ Are the sample size and selection method adequate?
☐ Are the validity and reliability of measurement instruments reported?
☐ Are the conclusions supported by the results?
☐ Are limitations discussed honestly?

Step 5: Synthesize

The most important principle — present a thematic, integrated narrative rather than a series of individual paper summaries.

Bad example: "Kim (2020) found A. Lee (2021) found B. Park (2022) found C."

Good example: "Studies reporting positive effects of online learning (Kim, 2020; Park, 2022) were primarily conducted in asynchronous learning environments, while the study reporting negative outcomes (Lee, 2021) identified a lack of interaction in synchronous settings as the primary cause."

Choose a synthesis approach — by theme, chronology, methodology, or findings — that best fits your research.

The Research Gap Finder categorizes gaps in the existing literature into four types during the synthesis process. This helps you build the logical thread of "synthesis → gap → positioning of the present study."

Step 6: Write the Review

Structure the literature review section as follows: introduction (scope and purpose of the review) → body (organized by theme, highlighting points of consensus, debate, and development) → synthesis discussion (patterns, gaps, implications) → conclusion (summary of key findings, justification for the present study).

NubintAI's AI Editor supports drafting with AI Autocomplete to continue your sentences, and the Citation Finder helps locate missing key references that you can insert directly in the editor.

Common Mistakes

Mistake	Solution
Using only one database	Search at least three
Confirmation-biased selection	Deliberately include studies with opposing findings
Just listing summaries	Use thematic synthesis to highlight patterns and flows
Focusing only on recent papers	Trace citation chains to include foundational works
Not documenting the search process	Record search date, query, and result count

Appropriate Scope for a Literature Review

Research Level	Recommended Number of Sources
Master's thesis	50–100
Doctoral dissertation	150–300
Journal article	30–60
Systematic review	Selected from the entire field

Summary

A literature review is the most time-consuming part of research, but it also determines the strength of your study's foundation. Thematic synthesis rather than paper-by-paper listing, documented search procedures, and pre-defined inclusion and exclusion criteria are the hallmarks of a strong literature review.