November 9, 2026

How to Screen Titles and Abstracts Consistently

Title-and-abstract screening is the stage where a literature search becomes a defensible body of evidence. You have a list of records, and you must decide, record by record, which ones move forward to full-text review. C

Written byWisPaper TeamAI Research Workflow Team
Editorial illustration for How to Screen Titles and Abstracts Consistently

Title-and-abstract screening is the stage where a literature search becomes a defensible body of evidence. You have a list of records, and you must decide, record by record, which ones move forward to full-text review. Consistency is the whole point: the same record should get the same decision regardless of when you read it, how tired you are, or which project you are working on.

This guide is for students and early-career researchers finishing a course paper, thesis chapter, or first manuscript and preparing the next project. It walks through the decisions and artifacts that make screening consistent, a complete sequence with checkpoints, a worked example, a reusable checklist, and clear completion criteria.

Why does screening drift, and what does consistency actually require?

Screening drifts for ordinary reasons. Eligibility criteria live in your head instead of on paper. You read records in a different order each session. You judge a borderline abstract differently on a Monday than on a Friday. You forget which records you already excluded and why.

Cochrane's guidance on searching and selecting studies points toward the fix: plan searches from explicit eligibility criteria, document the process, deduplicate records, and screen in stages. PRISMA 2020 reinforces the reporting side, providing flow-diagram templates and transparent reporting of search, selection, and synthesis methods. NIH rigor guidance frames careful assessment of prior knowledge, strengths, weaknesses, and knowledge gaps as core research practice.

Put together, consistency requires four artifacts before you screen a single record:

  • A written eligibility protocol with inclusion and exclusion criteria you can apply to a title and abstract alone.
  • A deduplicated record set so the same study is not judged twice.
  • A screening log that records each decision and the reason.
  • A staged process that separates title/abstract screening from full-text review.

What should your eligibility criteria look like before you start?

Criteria that only work at full text will fail at title/abstract stage. Write each criterion so a reasonable reader could apply it from the title and abstract alone, and mark the ones that genuinely require full text as "full-text only."

A workable structure separates three things:

  1. Population or topic — what the study must be about.
  2. Study type or design — what kinds of evidence count.
  3. Practical limits — language, date range, publication type, or setting, if you have justified them.

For each criterion, write one sentence explaining why it is there. That sentence is what you will consult when a record is borderline. If you cannot write the sentence, the criterion is probably not ready.

Cochrane's emphasis on planning searches from eligibility criteria is the reason to do this first: the criteria shape the search, and the search should not outrun the criteria.

How do you build and deduplicate the record set?

Cochrane recommends using multiple sources and combining free-text terms with subject headings. That produces overlap, so deduplication comes before screening.

A practical sequence:

  1. Run the search in your chosen database or scholarly search interface, saving the query and date.
  2. Export records with enough metadata to identify duplicates (title, authors, year, DOI).
  3. Deduplicate by DOI first, then by title and author match.
  4. Record the count before and after deduplication.

If you are working in WisPaper, the scholarly search interface is where you find academic papers, and My Library is where you save or upload papers and organize them. Keeping the deduplicated set in one library means every later decision refers to the same records.

Checkpoint: you should be able to state the number of records retrieved, the number after deduplication, and the date of the search. If you cannot, stop and fix the record set before screening.

What is the actual screening sequence, step by step?

Screening in stages is the Cochrane-recommended approach, and it maps cleanly onto a title/abstract pass followed by full-text review.

Step 1: Pilot the criteria on a small sample

Take 10–20 records and screen them against your written criteria. Note every record where you hesitated. Each hesitation points to a criterion that needs sharpening. Revise the criteria, then re-screen the pilot sample.

Step 2: Screen titles first, then abstracts

Title screening is fast and removes clearly irrelevant records. Abstract screening handles the genuinely ambiguous ones. Keeping the two passes separate makes your decisions easier to audit.

Step 3: Record a decision and a reason for every record

Every record gets one of three outcomes: include for full-text review, exclude, or mark as uncertain for a second look. The reason field should reference the criterion, not a free-form impression.

Step 4: Resolve uncertain records deliberately

Uncertain records are where consistency is won or lost. Batch them, re-read the criteria, and decide them together rather than scattered across sessions. If you are working with a supervisor or co-author, this is the natural point to compare decisions.

Cochrane notes the importance of linking multiple reports of one study. A conference abstract and a full paper on the same work should not be counted as two independent studies. Note the link in your log.

Step 6: Move included records to full-text review

Only now do you retrieve full texts. The title/abstract stage has done its job: it has produced a smaller, justified set.

Checkpoint after each step: the number of records in, the number out, and the reasons for exclusion should reconcile. If the counts do not add up, you have a record with no decision.

How do you keep decisions consistent across sessions and collaborators?

Consistency across time and people comes from artifacts, not memory.

  • Freeze the criteria. Once screening starts, changes to criteria should be dated and applied to the whole set, not just to new records.
  • Screen in batches. Decide a batch of records in one sitting against the same version of the criteria.
  • Use a decision vocabulary. Include, exclude, uncertain — and nothing else.
  • Keep the reason field short and criterion-linked. "Fails criterion 2 (not a primary study)" beats "not relevant."
  • Compare a sample with a collaborator. If two people screen the same 20 records, disagreements reveal ambiguous criteria, not careless colleagues.

If you are asking questions of the papers you have already collected, Library QA lets you ask questions based on papers uploaded to My Library or added from search results. That is useful for clarifying what a record actually reports before you decide on it.

Worked example: a January project start after winter-break cleanup

Suppose you are a master's student who spent winter break cleaning up a messy reference folder from last term. In January you start a new project on a specific topic, and your supervisor has asked for a short literature review by the end of the month.

Starting state: a folder of PDFs with inconsistent naming, no written criteria, and a vague sense that "some of these are relevant."

What you do:

  1. Write the protocol. Three inclusion criteria (topic match, primary study, within a stated date range) and two exclusion criteria (opinion pieces, duplicate reports). Each gets a one-sentence justification.
  2. Rebuild the record set. Run a fresh search, export records, deduplicate, and save the cleaned set in My Library.
  3. Pilot. Screen 15 records, note four hesitations, and sharpen two criteria.
  4. Screen. Title pass, then abstract pass, logging a decision and reason for every record.
  5. Batch the uncertain records. Decide them together against the revised criteria.
  6. Link duplicate reports. Note the conference abstract that matches a full paper.
  7. Move includes to full text. Retrieve and read them.

What "ready" looks like: a written protocol, a deduplicated set with counts, a log where every record has a decision and a reason, and a short list of included records ready for full-text review. You can hand that to your supervisor and they can see exactly how you got there.

If the project grows into a longer literature review, Scholar Agent supports longer research tasks in the literature review, inspiration, and reproduction areas. And if you are setting up the project itself, Research Projects is where you create and organize it.

Reusable screening checklist

Use this table as your working artifact. Copy it into your project notes and fill it in as you go.

StageArtifactCheckpoint question
ProtocolWritten inclusion/exclusion criteria with justificationsCan each criterion be applied from title and abstract alone?
SearchSaved query, source, and dateCan you reproduce the search?
DeduplicationCount before and afterDoes every remaining record have a unique identity?
PilotNotes on hesitations and revised criteriaDid the pilot change the criteria?
Title passDecision logDoes every record have a decision?
Abstract passDecision log with reasonsDoes every reason reference a criterion?
Uncertain batchResolved decisionsWere uncertain records decided together?
LinkingNotes on multiple reports of one studyAre duplicate reports linked?
HandoffIncluded listCan a collaborator follow your reasoning?

How do you know the screening result is ready?

Screening is ready when five conditions hold:

  1. The protocol is written and dated. Criteria exist on paper, with justifications.
  2. The record set is deduplicated and counted. You can state records retrieved and records after deduplication.
  3. Every record has a decision and a reason. No blanks, no free-form impressions.
  4. Uncertain records were resolved in a batch. They were decided against the same criteria version.
  5. Multiple reports of one study are linked. Your included count reflects studies, not reports.

If you are preparing to report the process, PRISMA 2020's flow-diagram templates give you a structure for showing records identified, screened, excluded, and included. Your screening log is the source for those numbers.

One more check: if you changed criteria mid-screening, confirm the change was applied to the whole set. A criteria change applied only to new records is the most common source of invisible inconsistency.

Where does citation checking fit?

Screening decides what enters your review. Citation checking decides whether the references you cite are real and correctly recorded. These are separate tasks, and both belong in a consistent workflow.

If you are assembling a BibTeX file from your included records, TrueCite checks BibTeX against real academic databases to flag hallucinated references. Running that check after screening, on the included set, keeps the two tasks from blurring together.

What is your next action?

Before you screen another record, write your eligibility protocol. Three inclusion criteria, two exclusion criteria, one justification sentence each. Then run a 15-record pilot and note every hesitation. That single hour converts a vague folder of papers into a screening process you can defend — and hands your supervisor something they can actually review.

FAQs

A pilot of roughly 10–20 records is enough to surface ambiguous criteria. The goal is not statistical representativeness but catching the hesitations that reveal a criterion needs sharpening. If the pilot produces no hesitations, your criteria may be too loose to discriminate.