How to Run a Journal Club That Actually Teaches Critical Reading

Every lab, graduate program, and research-methods course has a journal club, and most of them have the same problems. The presenter walks through the paper slide by slide while everyone else checks email. Almost no one reads the article beforehand, and increasingly, people skimmed an AI summary. Discussion, when it starts, ricochets between details: a statistical choice, a figure’s color scheme, a citation someone recognizes. The two most senior people talk. The meeting ends with no shared sense of whether the paper was any good, and the people who most needed to learn how to read critically have watched someone else do it instead.

This situation comes from a tradition and the absence of structure. Any good journal club, like an experiment, needs a design. This guide gives you one: a format that suopprts everyone’s participantion in critical reading, keeps the discussion on the argument of the paper rather than its trivia, and produces something useful at the end.

Four things every good journal club has

Before the format, the principles. Journal clubs that work share four features, and if yours is drifting, one of them is usually missing.

  • A clear goal. Is the group critiquing the methods, learning a new subfield, or deciding whether to build on the result? Each is a legitimate aim, and each produces a different discussion. Say which one you are doing.
  • Defined roles. One presenter and twelve listeners is a lecture. Distributing the work gives everyone a reason to prepare and a way to contribute.
  • Moderation. Someone has to keep time, bring quiet people in, and move the discussion forward when it stalls. This is a job, and it should rotate.
  • Preparation. If people arrive having skimmed the abstract, you get a summary of the abstract. The format below makes preparation short, specific, and checkable.

The shared framework: five elements, five questions

The format rests on a simple idea: every research article is built from the same five elements, so every discussion can follow the same five questions. What is the Claim? What Evidence supports it? Does the Reasoning get from the evidence to the claim? What are the Implications? What Context, what gap in prior work, motivated the study? That is the CERIC method, and its value in a group setting is that it gives everyone, from the first-year student to the PI, the same vocabulary for the same argument. The details of the field vary; the structure of the argument does not.

Before Or At the Beginning of the meeting

Every participant, not only the presenter, completes a one-page CERIC review of the article before or at the begininng of the meeting: one or two sentences per element, in their own words. The free cheat sheet is designed for exactly this, and the teaching pack includes a per-article analysis template. For an experienced reader this takes twenty minutes. For a newcomer it may take an hour the first time and much less after that, because the five questions are the same every week. It take about 10-12 hours of practice to feel proficient in the method, but then the gains are real and durable.

The presenter prepares the same review plus the figures, and template slides to be completed together as a group walkthrough. If the group has done its reading, the presenter’s job is to state the Claim in one sentence, add the consensus of which two or three figures that carry the evidence on screen, and add the consensus on reasoning. The presenter becomes a facilator who talks less and listens more.

During the meeting: a sixty-minute agenda

Times are for a one-hour session; scale them for longer or shorter meetings, but keep the proportions.

  1. Claim (5 minutes). Go around the room. Each person reads their one-sentence Claim aloud. They will not match, and that is the point. Where two readers disagree about what the paper is claiming, the paper is either making more than one claim or stating its claim unclearly, and either is worth knowing. Agree on a working version and write it where everyone can see it.
  2. Evidence (15 minutes). With the figures on screen, list the specific evidence for the agreed Claim. What was compared, how big the difference, how much uncertainty. Resist interpreting yet. The moderator’s job here is to keep saying “what does it show?” until the group has a plain-language list.
  3. Reasoning (15 minutes). Now interpret. How do the authors get from the list on the board to the Claim at the top of it? What has to be assumed? What alternatives did they rule out, and which did they not mention? This is the heart of the meeting and the part most journal clubs skip, because it is harder than either summarizing or nitpicking.
  4. Implications (10 minutes). If the Claim holds, what follows: for the field, for the group’s own work, for practice? Are the authors’ stated implications proportionate to their evidence?
  5. Context (10 minutes). What gap did the study set out to fill, and did the Claim fill it? Reading the introduction last, once the group knows what the paper found, makes this question answerable rather than rhetorical.
  6. Verdict (5 minutes). One sentence each: would you build on this result? The summarizer records the group’s consensus and the main dissent. That record is the output of the meeting, and it accumulates into something useful: a lab’s annotated map of its own literature.

Roles that rotate

Assign these each week and rotate them. Pairing a newcomer with an experienced member in each role works well.

  • Facilitator. Keeps the agenda and the clock, and brings in people who have not spoken.
  • Presenter. States the Claim, shows the figures, answers questions about the methods.
  • Skeptic. Comes prepared with the strongest case against the Reasoning. Having this as an assigned role makes criticism normal rather than personal.
  • Connector. Pulls the discussion back to the argument when it drifts (“How does that bear on the Claim?”) and links the paper to work the group knows.
  • Summarizer. Writes the one-paragraph record of the verdict and circulates it afterward.

Three common failures and the fix for each

Drift. The discussion wanders into interesting tangents and never returns. Fix: keep a visible “parking lot” list. Tangents go on it, the connector steers back, and the last five minutes can revisit anything on the list that still seems worth it. Curiosity is preserved; the analytical thread is not lost.

Detail ricochet. The group spends twenty minutes on a statistical choice in a supplementary table. Fix: the Evidence stage has a single question, “What does it show?”, and the moderator holds the group to it. Methodological disputes that matter will resurface in the Reasoning stage, where they belong, because that is where the question becomes “Does this choice affect whether the evidence supports the Claim?”

Dominance. The same two people carry every meeting. Fix: the round-robin Claim stage at the start guarantees everyone speaks in the first five minutes, which changes who speaks later. Rotating the skeptic role gives junior members safe harbor to criticize. And the pre-meeting written review means that a quiet participant’s reading is on paper even if it is not said aloud.

Running it online or asynchronously

The same structure works without a meeting at all. Using a social annotation tool (Hypothesis and Perusall are common), the group annotates the article in a shared layer during the week, tagging comments by element: #claim, #evidence, #reasoning, #implications, #context. The facilitator posts one prompt per element, and the “meeting” is the thread. In a study of an interdisciplinary graduate course that combined the CERIC method with social collaborative annotation in this way, students’ critical reading of primary literature improved measurably over four weeks (Bjorn, 2024, Frontiers in Education).

Two cautions from experience. Annotation groups tend toward over-highlighting, which buries the useful comments; a two-pass rule (first pass marks only the Claim, second pass only key Evidence) keeps the layer readable. And discussion threads drift just as meetings do; the same parking-lot tag works there too.

Resources for running it

The CERIC cheat sheet is the one-page pre-reading tool. The free teaching pack includes the analysis template, a discussion guide with twelve prompts organized by element, and a rubric if you are grading participation. The method itself, with chapters on using it for presentations, critique, literature review, and collaborative annotation, is in How to Critically Read the Scientific Research Literature (Cambridge University Press, 2026). If you want to launch a journal club or reading-intensive course with a live session for your group, see our workshops.

Related guides: How to Read a Scientific Paper and How to Evaluate the Evidence in a Research Article.