Switch course ▾
Session 4. What Works Here: Small Trials in Your School
Objective. Design a small, fair, ethical trial of one change in your school, collect simple data with a comparison, and read the results cautiously without overclaiming.
A small trial is a respectful way to find out what works with your pupils, but a careless trial can mislead you more than no trial at all. The key ideas gave you the design. This lesson looks at the details that make the difference: choosing a measure that you can trust and mark fairly, deciding in advance what you will do with each possible result, and handling the practical problems of real schools, such as absent pupils, a lost lesson or a colleague who is not enthusiastic.
Key ideas
Why a local trial
- Published evidence tells you what often works elsewhere. A local trial asks: did it work for my pupils, in my school, with my time?
- Keep it small and short: one change, 3 to 6 weeks, one clear measure. Level 2's action research (course 8.10) is the base; here you add a fair comparison.
- The result is a signal for your next decision, not a proof for a journal.
Design a fair comparison
- Best simple design: two similar classes (same grade, ideally the same teacher). One tries the change, the other continues as usual. Afterwards the second class gets the change too, so nobody misses out.
- Before you start, give both classes the same short test (baseline). Afterwards give the same test again. Compare the CHANGE in marks, not only the final marks.
- Weakest design: one class, before and after. Many things change over weeks, so write down what else changed (a holiday, a new teacher, a harder topic).
Read small numbers with care
- With 25 pupils a small gap can be luck: one absent pupil or one lucky guess changes the average. Look at the full list of marks side by side, not only the averages.
- Mark fairly: ask a colleague to mark the tests with pupil codes instead of names, so nobody knows which class is which.
- Words matter: say 'in our two classes the new routine was followed by a bigger gain', not 'the routine works'. Then repeat in the next unit; a result that repeats is worth more.
Ethics of trying things on pupils
- Never withhold something essential from the comparison class: they get your normal good teaching. Never try something you think could harm.
- Agree with your head of studies, follow school rules on informing parents, and keep marks anonymous with codes. Share results about groups, not about individual children.
- Keep the workload small. A trial that exhausts you will not be repeated.
A small trial as a cycle
A measure you can trust
Choose one short measure that matches what you changed. If you change how pupils explain fractions, test explaining fractions, with the same questions and the same marking guide before and after. A task that pupils can finish in 15 minutes is better than a long test you will not mark carefully. Consider a second, broader measure as a check, such as a task from the end of the unit, but keep it to one or two measures so that you do not hunt for a good result.
Mark fairly. Replace names with codes, ask a colleague to mark, and mark all scripts of one question at a time. Include every pupil who took part in both tests; if some were absent, say so and show the numbers with and without them. Never choose which pupils to count after you have seen the marks.
Decide first, then look
Before the trial, write what you will do for each result: 'If the gain is clearly bigger and the lists show it, we repeat in the next unit. If there is no difference, we keep or drop depending on the time cost. If the comparison class gains more, we look at why.' This stops hope from deciding afterwards. It also makes it easier to accept a disappointing result, because you chose the rule while calm.
Record what else happens: a lost week, a visit, a change of seating. Keep a diary of two lines a week. When the result arrives, read it alongside the diary. If the pattern holds in a second class pair or a second unit, your confidence grows; if it does not, you have learned something useful and cheap.
Examples from the classroom
Swapping in Samarkand
Scene. Jasur wants to test short written reflections at the end of lessons in Grade 6 science. He has two parallel classes. 6A does the reflections for four weeks while 6B continues as usual, and both write the same five-question check before and after. After four weeks he gives 6B the reflections for the next unit, so that no class misses out. A colleague marks the checks using codes.
Insight. A swap keeps the comparison fair to pupils as well as to the evidence.
A different starting point
Scene. In Tashkent, Malika finds that class 7A started one mark above 7B at baseline. At the end of her trial 7A is still one mark above. Before she says anything about the method, she compares the change in each class: both gained about the same. She tells the methodical association that the trial shows no clear benefit in these classes and that she will repeat it with the new unit before deciding.
Insight. Compare the change from baseline, not only the end marks.
A result of no difference
Scene. Dilnoza in Bukhara tested pair talk before writing for three weeks. The lists for both classes look the same. She is disappointed but writes an honest one-page note and shares it. A colleague thanks her: she had been about to buy a set of cards for the same idea. The school decides to park it, record why, and revisit it when the Grade 4 teachers have tried a different version.
Insight. A clear 'no difference' result saves time and money, and sharing it is useful.
Common pitfalls
- Changing two things at once, so that nobody can tell which one mattered.
- Using different tests before and after, or marking the two classes by different people with different standards.
- Dropping absent or low-scoring pupils from the results after seeing them.
- Treating a first small result as proof and spreading it to the whole school.
Plan a Trial and Read Its Results
- Pairs (20 min). Open the Activity Kit: Part A is Aziz's question in Bukhara: 'Does a 5-minute quiz on last lesson's ideas at the start of the lesson help my Grade 7 pupils?' Complete the trial plan template for him.
- Check your plan against five points: one change only; a comparison; the same test before and after; marked by someone who does not know the groups; a decision rule written BEFORE (for example 'if the gain is bigger and the lists show it, we repeat next unit').
- Swap with another pair. They play the doubting colleague and ask: what else could explain a difference? Improve the plan (10 min).
- Second activity (20 min). Part B gives SAMPLE results from Aziz's two classes (8 pupils shown each, for practice). Compare baseline and final lists, write a three-sentence cautious conclusion, and name one thing you would check or repeat.
- Whole group (10 min): read conclusions aloud. Which words were careful? Which went too far?
Your facilitator will guide you through this activity in the session.
Questions to discuss
- What was the easiest way a plan could go wrong?
- Why did we compare the change in marks, not only the final marks?
- What would you tell a pupil's parent who asks why their child's class is trying something new?
Reflect
- What change have you made in the past that you never checked? What could you have compared?
- Which colleague could be your comparison class partner?
- How will you feel if the result shows no difference? How will you share it?
Try it tomorrow
- Write the decision rule for your trial on the plan sheet before you start.
- Keep a two-line weekly diary of what else changes while the trial runs.
- Agree with a colleague who will mark the post-test using codes.
Further reading and links
Change one thing, compare fairly, decide your rule first, read small numbers with care, and treat every pupil with respect.