ACADEMIC WRITING GUIDE
How to Conduct a Systematic Literature Search for a Dissertation (With Worked Examples)
A literature search can look productive while quietly missing important studies. You may find dozens of articles, yet be unable to explain why you selected them, which databases you searched, or whether your search could be repeated. A strong dissertation search is more than typing a topic into Google Scholar: it is a planned, documented process that connects the research question to the evidence actually retrieved.
This guide shows how to develop a reproducible search strategy, adapt it for academic databases, screen records, and describe what you did. It includes three hypothetical worked examples and an illustrative screening flow. The examples are teaching models—not real database searches or evidence of actual research gaps.
What makes a literature search systematic?
A systematic search uses an explicit question, planned sources, transparent search terms, documented decisions, and a traceable record of what was found. It aims to reduce arbitrary selection. That does not automatically mean your dissertation is a formal systematic review: a conventional literature review can use structured searching without claiming the full methods or reporting standards of a systematic review.
The review type determines how comprehensive the search must be. A structured search for an ordinary dissertation literature review can be transparent and defensible without claiming the full methods of a systematic review. A formal systematic review generally also requires a prespecified protocol, suitably comprehensive source coverage, reproducible strategies, rigorous screening and critical appraisal. Scoping reviews map available evidence and may follow JBI guidance. Describe the methodology actually performed—not a label that sounds more impressive.
Start by writing a focused question. Our research-question guide explains how to narrow a broad interest into concepts that a search can actually investigate.
Step 1: Turn your research question into searchable concepts
Take the proposed question: “How does formative written feedback relate to undergraduate academic engagement?” The two central concepts are written feedback and academic engagement. The population—undergraduate students—may be a third concept, depending on whether including it in every database query excludes useful studies. Avoid searching every word of the research question as one literal phrase.
For intervention-focused questions, a PICO framework (population, intervention, comparison, outcome) can help; for exploratory scoping questions, Population–Concept–Context (PCC) may be more suitable. Do not force a framework onto a question it cannot represent.
| Concept | Possible search terms | Why included |
|---|---|---|
| Written feedback | “written feedback”, “formative feedback”, “assessment feedback” | Captures synonyms for the educational process |
| Engagement | “student engagement”, “academic engagement”, “learning engagement” | Captures the outcome or experience |
| Population (optional) | undergraduate*, “university student*”, “higher education” | Narrows to a study setting when justified |
Step 2: Build a search string with Boolean operators
Use OR to connect alternatives within one concept; use AND to combine separate concepts. Quotation marks often request a phrase, while an asterisk may expand word endings. However, quotation, truncation, field tags, and proximity rules differ across databases. Test the syntax in each interface rather than assuming that one string works everywhere.
Worked search string — educational feedback
("written feedback" OR "formative feedback" OR "assessment feedback") AND ("student engagement" OR "academic engagement" OR "learning engagement")
This broad first pass looks for studies containing at least one feedback expression and at least one engagement expression. Adding AND (undergraduate* OR "university student*") can improve relevance but may exclude studies that use terms such as “higher education learners.” Test both versions against a small set of known relevant papers before choosing one.
Wrong approach: feedback AND students AND university AND engagement AND grades AND motivation. Every additional AND can narrow the search so aggressively that studies relevant to the actual question disappear. More terms are not necessarily a stronger search.
Step 3: Choose databases that match the discipline
Choose sources because of their subject coverage, not simply because you can access them. Education questions might warrant ERIC plus a multidisciplinary citation database. Health questions may require MEDLINE/PubMed and subject-specific sources; business questions may use ABI/INFORM or Business Source where available. Record the platform as well as the database, because the same database can behave differently through different interfaces.
Google Scholar can support exploratory searching and backward or forward citation chasing. For a formal review, however, its variable ranking, retrieval and reporting features limit its value as a sole reproducible source. Database choice should reflect disciplinary coverage, access and the review question; consult a research librarian where possible.
The search is one part of a wider methodological decision. Our research-methodology guide explains why the chosen design must follow the question rather than convenience.
A platform-specific illustration: in PubMed, ("appointment reminders"[tiab] OR "text messaging"[tiab]) AND (outpatient*[tiab] OR clinic*[tiab]) targets title/abstract words. This is only a teaching example, not a validated complete health-review strategy. PubMed’s [tiab] field tag and wildcard syntax must not be copied verbatim into ERIC or Scopus: inspect each service’s help pages and translate the query before running it.
Step 4: Pilot the strategy before the full search
Pilot the strategy against several known relevant papers (“seed records”), inspect both relevant and irrelevant results, and adjust synonyms or field restrictions that create avoidable omissions. Record each version and why it changed. JBI describes a three-stage approach for evidence syntheses: an initial exploratory search, a comprehensive translated database search, and supplementary searching such as citation chasing and appropriate grey literature. The exact requirements depend on the review type.
For formal evidence syntheses, consider an independent search-strategy review. The PRESS checklist examines translation of the question, Boolean/proximity operators, subject headings, text words, spelling, syntax, and limits. A plausible search string can still miss studies because a field tag or bracket is wrong; peer review helps catch those failures.
Before committing to the final search, ask: Does each search concept belong in the question? Can the strategy retrieve a few known relevant studies? Are OR groups bracketed correctly? Do exclusions or date limits eliminate important evidence? Does the database interpret the symbols as intended? Document corrections rather than silently replacing a query after screening.
Step 5: Define eligibility rules before screening
Specify the study designs, populations, settings, phenomena or outcomes, date range, languages, and publication types relevant to the question. Explain why each restriction is defensible. A five-year date filter may be suitable for fast-changing technology but could unnecessarily exclude foundational theoretical studies. Do not introduce criteria after reading results merely to exclude inconvenient evidence.
| Decision | Illustrative rule | Reason |
|---|---|---|
| Include | Empirical higher-education research examining written feedback and student engagement | Addresses both core concepts |
| Exclude | School-level studies with no higher-education subgroup | Outside the population |
| Include or flag separately | Conceptual papers explaining feedback engagement mechanisms | Useful for background, but not necessarily empirical synthesis |
| Not automatically exclude | Non-significant or contradictory findings | Excluding them can bias the review |
Eligibility must align with the study boundaries. The scope-and-delimitations guide shows how to justify restrictions without confusing them with weaknesses discovered later.
Step 6: Record each search so that someone else can follow it
For every database, record the provider and platform (for example, MEDLINE via PubMed versus MEDLINE via Ovid), full search exactly as executed, date, filters, coverage, hit count, and export method. The same database may support different syntax through different platforms. Save search histories or screenshots and version the log after any material change. PRISMA-S provides a 16-item reporting checklist for searches used in systematic reviews.
| Database and platform | Date searched | Search details | Records |
|---|---|---|---|
| ERIC (example only) | 14 Oct 2026 | Feedback AND engagement concepts; title/abstract subject fields as supported | 84 |
| Scopus (example only) | 14 Oct 2026 | Translated keyword string; no language filter | 126 |
| PsycINFO (example only) | 14 Oct 2026 | Platform-specific subject terms plus free text | 73 |
The counts in this table are invented solely to illustrate a search log. They are not results retrieved by Acadelyra or evidence of how many relevant studies exist.
Step 7: Export, deduplicate, and screen transparently
Export records with identifiers, titles, abstracts, and citation details into an approved reference or screening tool. Remove duplicates carefully: the same paper may appear in several databases with slightly different titles or missing DOIs. Preserve a deduplication log rather than deleting records irreversibly.
Then screen titles and abstracts against the predetermined eligibility criteria. Retrieve full texts for potentially relevant studies and record the specific reason each full text is excluded. In a formal review, your methods may require more than one independent screener and a disagreement-resolution procedure. Follow the protocol; do not assume a single-reviewer approach meets every review standard.
Worked screening-flow calculation
Suppose a hypothetical search retrieves 84 ERIC records, 126 Scopus records, and 73 PsycINFO records. Total records identified = 84 + 126 + 73 = 283. After removing 58 duplicate records, 225 unique records remain for screening. If 170 are excluded at title/abstract stage, 55 reports are sought. Suppose 5 cannot be retrieved, leaving 50 full texts assessed. If 34 full texts are excluded with recorded reasons, 16 studies remain.
Arithmetic check: 283 − 58 = 225; 225 − 170 = 55 records sought for retrieval; five reports not retrieved leave 50 reports assessed; 34 full-text reports excluded leave 16 reports included. In this teaching example, assume each included report describes one distinct study; otherwise the final study count could differ. PRISMA distinguishes records (database entries), reports (documents), and studies (investigations). These counts are hypothetical, not measured results.
PRISMA 2020 provides flow-diagram templates and reporting items for systematic reviews. Do not add a PRISMA figure simply to make an ordinary narrative literature review look systematic.
In a real review, do not assume every exclusion belongs at the same stage: remove duplicates before screening, count title/abstract exclusions separately, document reports not retrieved, and give specific full-text exclusion reasons (for example, wrong population or study design). If two reports describe one study, retain the relationship rather than counting two separate studies.
Step 8: Extract study information and assess quality appropriately
Build an extraction form before you begin synthesising. Fields may include citation, research design, participants, context, key measures or themes, findings, limitations, and notes on relevance. Pilot the form with a few different studies to see whether it captures what your question requires.
Formal systematic reviews generally need a fit-for-purpose critical-appraisal or risk-of-bias approach. Different designs require different instruments; a qualitative interview study should not automatically be judged using a checklist designed for randomised trials. A scoping review has a different purpose, and critical appraisal is not universally required. State clearly which assessments you performed, why, and how they shaped the interpretation.
Once eligible studies are identified, the next challenge is synthesis. Our academic-source synthesis guide explains how to compare evidence rather than arrange a paragraph around each source in turn.
Step 9: Write an accurate search-methods paragraph
Your final report should answer: What question guided the search? Which sources and platforms were used? When were they searched? What search concepts and limits were applied? How were records deduplicated and screened? Where can readers access the full search strings? Never report database names or screening counts that you did not actually use.
Worked example 1 — Higher-education feedback
Hypothetical question: How is perceived usefulness of written feedback associated with engagement among university students? Planned sources: ERIC and PsycINFO, supplemented by backward citation searching. The pilot string groups feedback synonyms with engagement terms. Undergraduate-related terms would be tested before introducing them as restrictions. Eligible empirical studies would be defined in advance, with conceptual sources retained for background but separated from empirical synthesis.
Model methods paragraph (replace all fictional details before use): “A structured literature search was planned for ERIC and PsycINFO using terms for written or formative feedback combined with terms for student or academic engagement. Searches would be translated to the syntax of each platform and piloted against known relevant studies. Eligible empirical studies would address higher-education participants and both concepts. Records would be exported, duplicates reconciled, and potentially eligible full texts assessed against predefined criteria; the full search strings, dates, filters, and screening counts would be reported in an appendix.”
Why it works: It specifies the information sources, central concepts, eligibility basis, and planned screening process. Because it is a proposal, the verbs do not falsely claim the searches have already happened. After completing the work, report the actual databases, dates, totals, and decisions in past tense.
Worked example 2 — Clinic appointment reminders
Hypothetical question: What evidence exists about digital appointment reminders and non-attendance in outpatient care? A review focused on effectiveness would define the intervention, comparison, and attendance outcome and identify appropriate study designs. The search could pair reminder-related terms (SMS, text message, mobile notification) with outpatient appointment and attendance terms. Depending on review scope, MEDLINE and other health databases may be appropriate; database selection and full strategies would require review.
Illustrative search concepts: ("text message*" OR SMS OR "appointment reminder*") AND (outpatient* OR "clinic appointment*") AND (attendance OR "no-show*" OR "missed appointment*"). This is a conceptual example, not a verified PubMed query. Actual database tags, controlled vocabulary, wildcard handling, and hyphenated terms must be adapted and tested.
Model paragraph: “The planned search will combine controlled vocabulary and free-text terms describing digital appointment reminders, outpatient settings, and attendance outcomes in selected health databases. Eligibility criteria will define the target setting, intervention, comparison, outcomes, and study designs before screening. Search dates, translated database strings, records retrieved, deduplication and exclusion reasons will be retained to support transparent reporting.”
Why it works: The method follows a narrow effectiveness-oriented question and makes eligibility decisions explicit. It does not claim reminders reduce missed visits, and it does not invent review results.
Worked example 3 — Remote work experiences
Hypothetical question: How do early-career employees describe belonging in remote teams? An exploratory qualitative evidence synthesis may require broader experiential synonyms, such as identity, inclusion, social connection, or sense of belonging. Suitable business, psychology, and multidisciplinary databases should be selected with librarian input. A design-only filter for experiments would be inappropriate if the aim is employee experience.
Illustrative concept string: ("remote work" OR telework* OR "virtual team*") AND (belonging OR inclusion OR "social connection") AND (employee* OR worker*). Inspect retrieved qualitative studies before deciding whether to add interview or ethnograph* terms. Searching only “belonging” may miss relevant research framed as organisational identification.
Model paragraph: “The search strategy will combine terms for remote-work contexts with terms related to belonging and social connection. After an initial pilot search, the terms will be refined to capture relevant qualitative accounts without excluding studies that use different theoretical labels. Database-specific searches, citation searching, eligibility decisions and extraction procedures will be documented to support a transparent synthesis.”
Why it works: It demonstrates how search concepts respond to a qualitative question and acknowledges differences in terminology rather than over-restricting the search. The paragraph intentionally does not claim that any particular studies were found.
Common mistakes that weaken dissertation searches
-
Searching only one phrase and assuming relevant studies must use your wording.
-
Using OR between unrelated concepts or adding too many AND terms without pilot testing.
-
Copying one database query unchanged into another database with different field tags.
-
Applying unexplained language or publication-date restrictions.
-
Treating Google Scholar search ranking as a transparent or fully reproducible systematic search.
-
Counting database hits as included studies before deduplication and screening.
-
Failing to preserve the exact searches and full-text exclusion reasons.
-
Claiming PRISMA compliance for a literature review that did not follow formal systematic methods.
-
Selectively excluding studies because their findings contradict the preferred argument.
Frequently asked questions
Can I conduct a systematic search without writing a systematic review?
Yes. A narrative or dissertation literature review can use a structured, transparent search. Describe the methods you actually used and avoid labelling the entire review a systematic review unless its methodology justifies the term.
How many databases should I search?
There is no universal number. Choose sources that cover the research question and meet your institution’s expectations. A formal review may require several complementary databases and additional searching methods.
Should I use Google Scholar?
It can help with citation chasing and finding accessible records, but its search interface and ranking are not a substitute for fully reported database strategies when comprehensive systematic searching is required.
What is the difference between PRISMA and PRISMA-S?
PRISMA 2020 guides reporting of systematic reviews broadly. PRISMA-S focuses specifically on reporting search methods and search strategies; neither checklist automatically guarantees a good search.
Can I exclude older studies?
Only with a defensible reason related to the question. Recent-only limits can miss foundational evidence and distort conclusions.
What if I cannot access a full-text paper?
Record the attempt, use legitimate institutional access or interlibrary loan when available, and transparently report any reports not retrieved. Do not silently count an inaccessible paper as screened and excluded for substantive reasons.
Final search-and-reporting checklist
-
A clear review question and suitable review type are documented.
-
Main concepts, synonyms, and any controlled vocabulary are identified.
-
Boolean logic, quotes, wildcards, and database syntax have been tested.
-
Information sources were selected for their disciplinary coverage.
-
All dates, platforms, full strings, and justified limits are saved.
-
Eligibility rules were set before screening and applied consistently.
-
Deduplication, screening, and exclusion records are traceable.
-
Record/report/study counts have been reconciled and checked.
-
Extraction and appraisal plans suit the research question and designs.
-
No fictional counts, database searches, or review findings are presented as real.
-
Claims about PRISMA or other reporting guidelines match work actually completed.
-
The final report explains strengths and search limitations honestly.
Final takeaway
A strong literature search is not the one that finds the largest number of papers. It is the one whose sources, terms, boundaries, records, and decisions can be explained and defended. Start from the question, search deliberately, document each step, and present what the evidence shows—including uncertainty and conflicting findings. The result is a stronger foundation for the dissertation rather than a longer reference list.
Need support reviewing your search plan or structuring a transparent dissertation methods section? Acadelyra can help you understand the steps, organise your evidence and refine your academic writing while keeping the research decisions and source verification in your hands.
Academic and methodological references
• Cochrane Handbook, Chapter 4: Searching for and selecting studies
• PRISMA-S reporting guideline (EQUATOR Network)
• PRISMA 2020 statement and flow diagrams
• JBI Manual: Scoping reviews compared with other review types
• PRESS 2015 search strategy peer-review checklist (Karolinska Institutet)
• PubMed Help: Boolean operators, field tags and wildcard syntax (NLM)
Editorial note: This guide offers general academic-writing instruction. Formal systematic or scoping reviews should follow the applicable institutional, disciplinary and reporting requirements. Every worked example and search count above is hypothetical.