ACADEMIC WRITING GUIDE
How to Write Inclusion and Exclusion Criteria for a Systematic Review Dissertation (With Worked Examples)
A systematic review can be meticulously searched and still reach an unreliable conclusion if its authors make inconsistent decisions about which studies belong in it. One reviewer may include studies of adult university students while another includes school pupils. A paper about an online feedback platform may be treated as eligible even though the review question concerns written instructor feedback. A study may be discarded because it reports an inconvenient result—even though its design and population meet the original protocol.
Clear inclusion and exclusion criteria for a systematic review make these decisions explicit before researchers see the pattern of results. They define which evidence can answer the review question, protect against arbitrary study selection, and make the final dissertation reproducible. This article shows how to develop, justify, pilot, apply and report eligibility criteria, including a fictional worked screening example, a decision log and model dissertation paragraphs.
It focuses principally on intervention-focused systematic reviews. Different review designs—including qualitative evidence syntheses, diagnostic test reviews and scoping reviews—require questions and eligibility frameworks suited to their aims. The underlying principle remains the same: eligibility rules should follow the question and method, not the outcomes researchers hope to find. The main methodological anchors here are Cochrane Handbook Chapter 3 and PRISMA 2020.
Teaching-example disclaimer: All fictional studies, institutions, interventions, dates and screening decisions in the examples below are invented solely to explain the method. They are not real research evidence. Do not present these examples as studies you found or screened.
What are inclusion and exclusion criteria in a systematic review?
Inclusion criteria specify characteristics that a study must have to be eligible for the review. Exclusion criteria specify characteristics that make a study ineligible. Together they describe the review's boundaries: eligible populations, exposures or interventions, comparators, outcomes, designs, settings, reporting characteristics and other predeclared restrictions.
For an intervention review, the PICO framework—Population, Intervention, Comparator and Outcomes—is a useful starting point. However, PICO does not replace decisions about study design, follow-up, language, dates and whether multiple reports describe the same underlying study. The Cochrane Handbook's scope and question guidance distinguishes the review-level PICO from the questions of individual syntheses; this matters when different eligible comparisons are reported separately.
A broad criterion such as “relevant peer-reviewed articles” is not reproducible. A reader cannot independently decide what the reviewer considered relevant. A better criterion states a characteristic that can be observed from a study report and linked to the research question.
Eligibility is also different from quality appraisal. A study can meet the inclusion criteria but have serious risk-of-bias concerns. Ordinarily, risk of bias is assessed and interpreted using an appropriate appraisal method; it should not become an improvised reason to drop inconvenient studies. If a review protocol genuinely limits eligibility to specified designs or methodological features, that decision needs advance justification and transparent reporting. See our guide to assessing risk of bias in a systematic review dissertation.
Start with a focused review question
Consider this fictional dissertation question:
Among first-year undergraduate students, does structured written formative feedback from instructors, compared with usual feedback practice, improve course completion during the first academic term?
This question contains several important boundaries. The participants must be first-year undergraduates; the intervention must be a structured form of instructor-written formative feedback; the comparison must be ordinary feedback practice; and the key outcome is course completion during the first term. A paper about general student satisfaction may be informative background reading, but it may not answer this review question.
The distinction between eligibility for the review and availability for a particular synthesis is crucial. A study may meet the review's population, intervention, comparator and design criteria but lack a usable measure for the primary synthesis. Depending on the prespecified protocol, it may still be included in the review and described transparently, rather than being discarded simply because its statistics cannot be pooled. Plan these cases in advance with guidance from Cochrane Chapter 3.
Translate the question into testable rules
Begin with a provisional eligibility table. Each criterion should tell screeners what is eligible, what is not and what evidence would resolve uncertainty.
| Domain | Include | Exclude or flag for clarification |
|---|---|---|
| Population | First-year undergraduate students; mixed-year samples with separately reported eligible first-year results | School pupils, postgraduate-only samples or mixed cohorts with no identifiable eligible group when the protocol requires one |
| Intervention | Instructor-delivered structured written formative feedback linked to assessed work | Automated grade notification alone; verbal-only feedback; broad teaching redesign without identifiable written-feedback component |
| Comparator | Usual instructor feedback or another prespecified comparison | No eligible comparator when the review is restricted to comparative effects |
| Outcome | Course/module completion as defined in the protocol, with eligible timing | Studies collecting only perceptions where the protocol requires a completion outcome; distinguish absent reporting from demonstrated absence of measurement |
| Study design | Prespecified randomized and eligible non-randomized comparative designs | Editorials, commentary and non-comparative descriptions unless explicitly eligible under a different review question |
| Setting | Credit-bearing undergraduate courses | Non-degree workshops or informal training programs outside the target setting |
| Report status | Published studies and eligible unpublished reports, if these are sought by the protocol | Sources outside prespecified report-status criteria; do not exclude solely because a result is non-significant |
| Language and dates | As defined and justified in the protocol | Any restriction must be prespecified and reported; do not impose a convenient cutoff after seeing search results |
This matrix is an example of decision structure, not a recommended universal set of restrictions. Whether a study with no reported completion data belongs in the overall review depends on whether the review's eligibility is defined by the outcome being measured, reported, or something else; Cochrane cautions about selection based on available outcome results. The dissertation's own protocol must resolve that distinction appropriately.
Eight decisions to make before screening
1. Define the population without hiding borderline cases
Specify relevant ages, educational levels, conditions or settings. Avoid loosely using labels such as “students” if your question concerns undergraduates at a particular stage. If a sample includes first-year and later-year students, define whether subgroup data, a minimum relevant proportion or an author-confirmed eligible subset is required. Do not invent a subgroup rule only after finding a desirable study.
Think about the unit of eligibility. A study may include several participant groups, yet only a subset may answer your question. Record which data can be used and whether the mixed sample prevents clear inference. A narrow inclusion rule might improve applicability but reduce available evidence; a broad rule might increase heterogeneity and weaken the specificity of conclusions.
2. Describe the intervention or exposure operationally
Labels are not enough. “Digital feedback” might mean instructor comments, automated quiz scores, peer comments or a complete learning-management intervention. Define the minimum features that make a study's intervention eligible, including source, format, timing and core function where relevant.
If the review includes multicomponent interventions, state whether an intervention must isolate the component of interest or whether combined programs are eligible. Decide how such studies will be grouped for interpretation. Failing to do so may lead the review to attribute an effect to feedback that could have arisen from several bundled changes.
3. Specify eligible comparison groups
For a question about comparative effectiveness, a comparator is not an administrative detail. State whether usual practice, no feedback, alternative feedback methods or other active comparators qualify. When comparisons differ substantially, plan separate synthesis groups rather than combining them merely because they share the intervention's name.
4. Define outcomes and time windows carefully
A study's outcome can be measured, reported or extractable in different ways. Make clear which outcomes the review seeks and which timing is relevant. Course completion, attendance and engagement are related but not interchangeable. If a paper reports an eligible outcome at several follow-ups, the rule for choosing results for a particular synthesis should be decided before examining which estimate looks strongest.
Critically, do not exclude an otherwise eligible study merely because a usable result for the outcome of interest is missing, unless the review's prespecified and justified eligibility definition truly requires the outcome to have been assessed. Distinguish review inclusion, outcome-data availability and synthesis inclusion; record missing results transparently. Our data extraction form guide explains how to track these distinctions.
5. Choose study designs that match the inference
Randomized trials may be useful for intervention-effect questions, but they are not the only possible eligible studies. Depending on the question, quasi-experimental comparisons or other non-randomized designs may be appropriate. Explain the rationale and plan for different bias structures and synthesis groups. Do not classify every observational comparison as a randomized trial or quietly add excluded designs when too few articles are found.
6. Decide which report types count
A study and a report are not synonymous. One study can have a journal article, a thesis, a registry record and a conference report. Multiple reports may supply complementary information while describing the same participants. Keep stable study identifiers and linked report identifiers so the same sample is not double-counted. The PRISMA 2020 explanation distinguishes records, reports and studies for this reason.
Specify whether unpublished studies, preprints, dissertations, trial registers and conference abstracts are eligible. Excluding all non-journal evidence without considering the question can create a publication-status limitation; including every format without checking study identity can create duplicates.
7. Justify dates, languages and geography
A date cutoff can make sense when technology or policy changed substantially, but “only recent papers” is not enough on its own. State the relevant date type: study conduct, database coverage or publication. Similarly, language and country restrictions should reflect the question or a disclosed practical constraint rather than be hidden by the search strategy. Report likely limitations of such restrictions.
8. Plan for unclear abstracts and inaccessible full texts
A title or abstract may not contain enough information for exclusion. The screening protocol should explain when an uncertain record advances to full-text review and how unclear eligibility will be resolved. If a full text cannot be retrieved, record it as not retrieved, not automatically as “excluded for ineligible population.” The PRISMA 2020 flow guidance separates these situations.
How to apply the criteria consistently during screening
Eligibility rules become useful only when they are applied consistently. Cochrane's study selection guidance describes the steps from screening search records to deciding inclusion at full text. The Cochrane MECIR standards set expectations for independent eligibility decisions in Cochrane intervention reviews. A dissertation conducted by one researcher may face different institutional resources, but it must describe what was actually done—not claim duplicate independent screening that never occurred.
Pilot the decision rules on varied records
Before the main screening, try the draft rules on a purposively varied sample: an obviously eligible abstract, an obviously ineligible record, a mixed-population study, an unclear intervention and a duplicate report. Ask whether two reviewers using the same instructions would likely reach the same judgment. Document ambiguous cases and clarify wording prospectively.
A pilot is not permission to reshape eligibility around the observed effects. If a genuinely ambiguous rule requires revision, record its previous wording, new wording, reason, date and effect on records already screened. If necessary, revisit earlier decisions to maintain consistent application.
Use separate screening stages
In title/abstract screening, remove clearly ineligible records but retain uncertainty for fuller assessment according to the protocol. In full-text screening, use the complete report and linked information to make definitive eligibility decisions. Record the principal reason for excluding each assessed full-text report. Avoid mixing exclusion reasons with administrative issues such as inability to retrieve a report.
| Fictional report | Information available | Screening decision | Reason and next action |
|---|---|---|---|
| Alder 2024 | Randomized comparison of structured instructor comments in first-year courses | Include for study-level review | Eligible population, intervention and comparator; verify outcome details during extraction |
| Birch 2023 | Structured feedback intervention among secondary-school pupils | Exclude at abstract screening | Ineligible population under the stated protocol |
| Cedar 2022 | Undergraduate feedback study, year of study unclear in abstract | Seek full text | Do not guess that the sample is ineligible from incomplete abstract wording |
| Delta 2025 | First-year students receive only automated numeric score notifications | Exclude at full-text screening | Intervention lacks the prespecified instructor-written feedback component |
| Elm 2021 | Same trial and participants as Alder, in a follow-up publication | Link as additional report | One underlying study with multiple reports; do not count as a second study |
| Fir 2020 | Potentially eligible trial, but full report could not be obtained | Record as report not retrieved | Not equivalent to a confirmed full-text exclusion; document retrieval attempts |
These decisions are illustrative. A review that legitimately defines feedback more broadly or uses a different population rule would reach different eligibility decisions. What matters is fidelity to the stated protocol.
Record disagreements without editing history
If two reviewers disagree, retain both initial decisions, the criterion at issue, how the decision was resolved and the resulting inclusion status. If a supervisor checks a sample rather than screening everything independently, report that process exactly. An unexplained final “include” or “exclude” status does not show how ambiguity was resolved.
Use an exclusion taxonomy sufficiently detailed to explain decisions but not so granular that screeners cannot apply it consistently. Categories might include wrong population, wrong intervention, wrong comparison, wrong design and wrong study setting, with an optional note for the decisive detail. More than one reason may apply, but the review should use a consistent rule for recording its principal reason in the PRISMA documentation.
Build a transparent exclusion log
A screening spreadsheet or review software should enable another researcher to reconstruct decisions. Minimum fields include record/report ID, linked study ID, citation, screening stage, reviewer decision, uncertainty flag, full-text eligibility, primary exclusion reason, explanatory note, reviewer/verification information and date of decision. If your institutional process uses one researcher, do not invent second-reviewer entries.
| Field | Example entry | Why it matters |
|---|---|---|
| Report ID | REPORT-CEDAR-01 | Makes the exact publication traceable |
| Linked study ID | STUDY-CEDAR | Prevents counting different reports as distinct studies |
| Stage | Full-text assessment | Distinguishes definite decisions from abstract screening |
| Decision | Exclude | States final eligibility outcome |
| Primary reason | Wrong intervention | Connects the decision to a predefined criterion |
| Evidence location | Methods, intervention paragraph | Allows another reviewer to verify the exclusion |
| Uncertainty or resolution | Definition confirmed from supplementary report | Retains the context of borderline decisions |
| Decision date | Recorded at screening | Supports an audit trail for protocol and criteria changes |
The values above are examples of fields and formatting, not a record of genuine screening. If your review identifies several reports of the same study, keep the report-level exclusion record aligned with the study-level inclusion decision.
How to report eligibility criteria in a dissertation Methods chapter
A defensible Methods section should tell the reader what evidence was eligible, why the rules were chosen, when they were defined, who applied them and how ambiguous decisions were handled. PRISMA 2020 item 5 addresses eligibility criteria; item 8 addresses the selection process; and item 16 addresses the flow of records, reports and studies through selection.
Model Methods paragraph: eligibility criteria
Eligibility criteria were developed before screening to reflect the review's population, intervention, comparison, outcomes and eligible study designs. The review focused on [define population], [define intervention] and [define comparator], with [define outcome and timing] treated according to the prespecified protocol. Studies were eligible if they met [describe the precise design and setting rules]. Restrictions on language, date and report status were [state and justify actual rules]. Studies were excluded when they did not meet these criteria. Any clarifications or amendments were documented with their rationale and date.
This is a writing structure, not a statement that those procedures occurred. Fill the brackets with actual methods; report protocol changes honestly. For guidance on organizing the surrounding chapter, see how to write a dissertation methodology chapter.
Model Methods paragraph: study selection
After removing duplicate records, titles and abstracts were screened against the predefined eligibility criteria. Reports considered potentially eligible or unclear were sought for full-text assessment. Full texts were evaluated against the eligibility rules, and the principal reasons for exclusion were recorded. Screening was carried out by [accurately describe personnel and independence], and disagreements or unclear cases were resolved through [describe actual process]. Study IDs were used to link multiple reports from the same underlying study. The final selection process was documented in a PRISMA 2020 flow diagram.
Use this paragraph only if it reflects what truly happened. If there was no duplicate independent screening, no protocol, no documented disagreement procedure or no formal full-text log, do not imply otherwise. State the limitation and explain what was done.
Report study flow without confusing units
The final report should distinguish records found in searches, reports sought and assessed, and unique studies included. A conference abstract and journal article may both be reports of one study. A report that could not be retrieved is not the same as a retrieved report excluded because it fails an eligibility criterion. A reason for exclusion at full text should describe the actual mismatch with the rules. Our guide to creating a PRISMA 2020 flow diagram explains the flow categories and how to reconcile them.
Do not fabricate a diagram by selecting plausible counts. It must be built from the real search and selection log, with duplicate handling and any automation exclusions stated accurately.
Avoid these common eligibility mistakes
Excluding studies because they report non-significant results
Eligibility should not depend on whether the authors reported a statistically significant benefit. That would selectively remove contrary evidence and bias the review's conclusions. Use design and question-based criteria, then assess the eligible evidence transparently.
Writing restrictions only after seeing the literature
An arbitrary publication-year restriction or newly narrowed population rule can make a review appear cleaner while excluding relevant evidence. Define and justify rules before screening. If revision is unavoidable, document the change and assess whether it could influence results.
Confusing study inclusion with synthesis inclusion
A review may include an eligible study that lacks an effect estimate for a particular outcome synthesis. Record it appropriately and explain missing information. Do not quietly drop it from the entire study count merely because it cannot enter a pooled analysis. The distinction also matters for narrative synthesis reporting.
Treating all reports as separate studies
A trial registry, thesis and article may represent one study. Link them before calculating the number of unique included studies or interpreting the volume of evidence. The systematic review data extraction guide describes study/report identification.
Calling methods rigorous without showing what was done
Terms such as “systematic,” “independent” and “validated” are claims about procedures, not decorative descriptions. Support them with an actual protocol, decision records and honest screening methods. Follow the standards suitable for your review design and disclose departures from best practice.
Using software decisions without human verification
Screening tools can organize citations and flag likely duplicates, but errors occur. Record automation methods, verify material decisions and retain oversight of exclusions. A tool's confidence score is not itself a valid replacement for eligibility reasoning.
Frequently asked questions
Are inclusion criteria and exclusion criteria the same thing?
No. Inclusion criteria describe characteristics necessary to qualify for the review; exclusion criteria identify incompatible characteristics. A clear, jointly specified eligibility framework prevents contradictory screening decisions.
Should I include only peer-reviewed articles?
Not automatically. That restriction depends on the research question, review type, protocol and potential limitations. Eligible dissertations, trial registers, preprints or other reports may contain relevant evidence, though they need appropriate verification and appraisal.
Can I change eligibility criteria after screening starts?
Sometimes clarification or a justified amendment is necessary. Document the original and revised rules, timing, rationale and affected records; do not silently change criteria after seeing preferred results.
Is a study excluded if the abstract does not mention the primary outcome?
Not necessarily. Abstracts often omit details. Use the protocol's eligibility rules, seek full text for uncertain cases and distinguish absence of reporting from absence of outcome measurement. Avoid choosing studies based on availability of favorable outcome statistics.
How many exclusion reasons should I record?
Use a manageable, predefined set covering the key eligibility domains. At full-text screening, retain enough detail to identify the principal reason and permit audit. Keep the flow counts consistent with the report and study units.
Can one student screen all studies for a dissertation?
Institutional expectations and review methodology differ. Independent duplicate selection is recommended in Cochrane intervention reviews to reduce errors. If you screen alone, describe that truthfully and discuss the associated limitation; don't represent the process as equivalent to independent dual screening.
Do PRISMA flow diagrams establish that my eligibility criteria are correct?
No. PRISMA improves reporting transparency, not the methodological validity of each eligibility decision. A neat diagram can still conceal poorly justified restrictions or inconsistent screening unless its underlying methods are sound.
Are eligibility criteria the same as participant recruitment criteria?
No. In a systematic review, you select studies and reports relevant to a review question. In a primary research study, recruitment criteria select individual participants. The concepts are related, but their units and methodological implications differ.
Final checklist before approval
The question and review design identify the eligible population and phenomenon or intervention.
The comparison and outcome rules are appropriate to the question and do not selectively privilege favorable results.
Study designs and settings are defined precisely and justified.
Restrictions on language, dates, geography and report status are stated and defensible.
Study inclusion and outcome-specific synthesis eligibility are distinguished.
Multiple reports of one underlying study can be linked without double-counting.
Uncertain abstracts are not automatically treated as definite exclusions.
Full-text exclusions have clear, auditable reasons.
A change log captures any legitimate eligibility-rule amendments.
Reviewer roles and disagreement resolution are reported as actually performed.
The PRISMA flow distinguishes records, reports and unique studies.
No teaching example is presented as empirical dissertation data.
Final takeaway
Good inclusion and exclusion criteria do more than narrow a search. They connect the review question to a defensible body of evidence and make decisions visible to readers. State observable rules in advance, pilot their application, separate uncertain records from definite exclusions and preserve an audit trail. When a reviewer can explain why a report was included or excluded without looking at whether the result was favorable, the systematic review has a stronger foundation.
If you would like professional human support organizing eligibility matrices, reviewing the clarity of a Methods chapter or checking how a systematic review is reported, explore Acadelyra's Research Support and Dissertation & Thesis Support services. Acadelyra supports sound research practice and academic integrity; your review decisions, source evaluation and submitted scholarship must remain your own.
Authoritative references and further reading
Page, M. J., et al. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews.
Page, M. J., et al. (2021). PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews.
Cochrane. MECIR standards: Selecting studies to include in the review (C39–C42).