GAMSAT study group: a peer study protocol that stays on reasoning

A GAMSAT study group is worth running if its main job is letting each member hear how someone else reasoned through the same question. It stops being worth running when sessions turn into score swapping, file sharing, or settling questions about the examination by group opinion. The protocol below is built to keep the group to the first job. We have no outcome data on whether peer study changes scores, so we make no claim that it does. What we can offer is a structure that makes the time usable.

Finding a partner is the easy part

At least one commercial course lists study buddy matching among its features, with messaging between enrolled students and contact details kept private. Matching solves the problem of finding someone. It says nothing about what the two of you should do once you have met, and that second question decides whether the hour produces anything.

Compare reasoning moves, not scores

A score announces a result and hides the route to it. Two members could each get the same number right and have lost marks for unrelated reasons, and the number gives the group no way to see that. Announcing scores also sets up a ranking, which is a different activity from comparing reasoning. The question worth asking is what each person did at the step where they chose.

For a question each member picked themselves, ask the speaker to give four things, in this order:

• what the stimulus provided, summarised rather than read aloud;
• the step they took from the stimulus to their chosen option, and whether it was a reading of what was stated or an inference beyond it;
• what led them to set the other options aside;
• how sure they were before they checked.

ACER's booklet gives the group a shared vocabulary for Section 1. It sorts the emphasis of Section 1 questions under three headings: understanding, which takes in meanings stated and meanings implied; plausible reasoning, which covers extending ideas and reaching conclusions; and critical thinking, which asks for discriminations and judgments within that plausible reasoning. Each question offers four alternatives and the candidate picks the most appropriate. When two members defend different options, ask which kind of step the question called for, and whether the step each of them took is one the passage supports at that level. An inference is not suspect for being an inference. The check is whether the passage licenses it, or whether something plausible was added.

Write your own practice items and keep protected material out

Passing practice material between members raises a rules question before it raises a study question. ACER's 2026 information booklet includes an entry on whether materials you have bought may be shared with other people, and the booklet is part of the agreement you enter on registering. Read its answer yourself before any file or screenshot moves between members. We do not summarise it here, because the wording is ACER's to give.

The protocol avoids the issue by not needing material to travel. Each member writes short items of their own: a paragraph with a few questions, or a small data table with a problem, built to exercise a type of reasoning they want to practise. The others attempt it and explain their routes. Do not rebuild real questions from memory, since an item that closely reproduces an examination question brings the same problem back in another form. Writing the item is useful in itself. The author has to decide what makes the intended answer better than the alternatives, and if the group cannot agree that an item has a best answer, the item becomes the thing under discussion.

Homemade items have a cost. Nobody in the group can know how closely they match the difficulty or format of the real thing, so use them to rehearse reasoning moves and not to judge readiness.

Treat peer feedback as a second reading, not a ruling

A peer can tell you where your reasoning was unclear to them and can offer a reading you had not considered. A peer cannot tell you what is true about the examination, and a group chat is a poor place to settle it. ACER states that important information is in its information booklet and on its website, and that anything unanswered goes by email to the GAMSAT office. Take how long results remain valid: the booklet says this varies by country and by institution, and tells registrants to confirm currency for the cycle they are entering. It refers eligibility questions to the institution you plan to apply to, or its admissions body. A member describing what their own university accepted is describing one institution. Deadlines work the same way. The booklet publishes them, and our guide to refund, deferral and change windows puts those windows in sequence.

A workable house rule: a statement about the examination, a deadline or an admissions requirement enters the group's notes only with the page it came from, whether ACER's or the institution's. Opinions stay in the conversation. The same rule covers theories about individual questions on test day, and our piece on trial questions looks at what ACER says about them.

Three meeting formats

Timed silent practice. Everyone works the same set under one shared clock, with no talking and no scores announced. Before anyone speaks, each person notes their confidence on every question and a one-line reason for the ones they were unsure about. Use only the working materials that ACER's booklet permits for the section; our article on scratch paper and whiteboards covers confirming those rules and deciding what the working space is for. A note written before discussion is a record. An account given afterwards has had the chance to absorb what other people said.

Reasoning reconstruction. Each member picks questions from their own materials, ones they got wrong or where confidence and answer disagreed, and explains the step they took in words that do not require the question text to leave their own materials. The others do two things only: say where they could not follow the step, and offer another reading. The speaker leaves with one thing to test on later questions. A single conversation does not show whether a pattern has changed, and what guided mistake review looks like describes a structured version you can repeat over time.

Section 2 clarity review. ACER's booklet describes Written Communication as two timed writing tasks of 30 minutes each, and every task presents four statements sharing one theme. Responses are assessed on the thinking behind them and on their organisation and expression. The booklet also says candidates are not assessed on whether the views they express are correct. That shapes what a reader should do. Each member writes alone and under time beforehand. The reader reports only where the line of thought became unclear, where they had to reread, and what other reading of the prompt the essay could be answering. The reader does not rewrite sentences, propose paragraphs, supply ideas or say whether they agree.

There is a rules-based reason to keep sentence-level help out. ACER says every response goes through plagiarism checks during marking, with AI use among the things checked and earlier GAMSAT submissions among the things compared against. It also says that a response prepared in advance scores low, and that text lifted from pre-prepared essays risks a flag even when the writer never saw the source. A group that circulates model paragraphs is producing the kind of shared wording that warning concerns. The booklet also answers a question about getting feedback on practice essays, which is worth reading before you design the feedback step. For what tools should and should not supply, see our piece on AI feedback without outsourcing your writing, and for what to take from model essays, sample essays without imitation.

Where this protocol falls short

A group gives you other people's accounts of their reasoning, and an account is not the same as what happened. A member who explains a step fluently may be describing a tidier version of it, and you cannot see which. A confident member can also become the group's authority simply by being confident, which defeats the second-reading rule, so rotate who speaks first. Self-written items are weaker proxies for the real examination, and scheduling costs hours. If comparison makes a session worse for you, a single study partner, or none, is a reasonable design. Nothing here is evidence that peer study raises a score.

What a group cannot give you is a record made at the time. Every account of reasoning is assembled afterwards, so nobody, you included, can easily see whether the same step is failing from week to week. Verantrix can record how you reasoned through a question, not only which option you chose, and can identify patterns across your attempts. Before the next meeting, write one line of reasoning and a confidence rating beside each question in your next practice set as you answer it, and bring those notes to the group instead of your score.