Switch course ▾
Session 2. Choosing Simple Evidence
Objective. Choose two or three simple, fair sources of evidence that fit your question, write a short evidence plan and avoid weak survey questions.
Evidence is not a mountain of paper. For a teacher it is usually a few numbers and a few sentences that are collected the same way every time. The danger is not too little evidence but the wrong kind: a feeling in place of a count, or a long survey nobody finishes. This lesson goes deeper than the key ideas, with a way to choose sources and a way to keep them fair.
Key ideas
Four everyday kinds of evidence
- OBSERVATION: a 5-minute tally or note of what pupils do, for example how many pupils answer aloud. EXIT TICKET: one to three questions in the last 3 minutes of a lesson.
- STUDENT WORK: a small sample of the same kind of task before and after. SHORT SURVEY: three to five tick-box questions, plus one open line.
- Tests and marks can be evidence too, but they come late and cover a lot at once. Use them as one source, not the only one.
Match the evidence to the question
- Ask: 'If my change works, what will change, and where will I see it?' Behaviour shows up in observation, understanding in exit tickets and work, feelings in a survey.
- Use at least two different sources. If a tally and an exit ticket point the same way, you can trust the finding more. This is called triangulation.
- Count the cost: if collecting takes more than about 5 minutes a day of your time, you will stop by week two. Choose the cheapest source that still answers the question.
Make it fair and consistent
- Collect the same way every time: same part of the lesson, same question, same counting rule. Write the rule before you start, for example 'a pupil counts once per lesson if they speak a full sentence'.
- Collect the baseline in the same way as the evidence during the change. Otherwise you compare two different things.
- Look at a small sample, not everything: six to eight pieces of work from different pupils is enough. Choosing only the best pieces gives a false picture.
Good survey questions and the log
- Write short questions with one idea each. Avoid leading questions ('Don't you agree...?'), double questions (two ideas in one) and vague ones ('Do you like it?').
- Make surveys anonymous, and let pupils skip a question. Honest answers need pupils to feel safe from judgement.
- Keep a one-line log per lesson: date, topic, the number, and anything unusual (a holiday, a test, many absent). You will need these notes in Session 4.
Three sources and what each shows best
Tally
Behaviour
- How many pupils answer aloud
- Cheap, inside the lesson
Exit ticket
Understanding
- 1 to 3 questions in 3 minutes
- Same wording each time
Survey
Feelings and views
- 3 to 5 tick boxes
- Anonymous, neutral words
Choosing by the question: behaviour, understanding, feelings
Ask what would change if your idea worked. If you expect pupils to do something more often, such as answering aloud or starting a task, a tally of that behaviour in a fixed part of the lesson is the closest evidence. If you expect them to understand more, use an exit ticket or the same kind of task before and after. If you expect them to feel more confident, a short anonymous survey is the best way to ask, although feelings can be mixed with a wish to please the teacher.
This is why two sources are better than one. A tally and an exit ticket look at different things. If both move the same way, you can trust the finding more. If they disagree, that is a finding too, and worth reporting honestly.
Fairness: the same way, a small sample, a written rule
The most useful habit is to write the counting rule before you begin. 'A pupil counts once per lesson if they say a full sentence aloud in the ten minutes after a question' is a rule; 'lots of pupils answered' is not. The rule keeps you honest on a tired Thursday and lets a colleague repeat your count.
Keep samples small and chosen by a rule. Looking at six notebooks picked by 'every fifth name in the register' is better than reading thirty at midnight or choosing the six that look best. Collect your baseline with exactly the same rule, at the same part of the lesson, so that 'before' and 'after' can be compared.
Examples from the classroom
Tally plus ticket in Bukhara
Scene. Jasur in Bukhara wants to know if two minutes of partner talk makes more Grade 5 pupils answer aloud in maths. He tallies pupils who say a full sentence in the first ten minutes after a question, and he ends each lesson with the line 'I can explain today's idea to a friend: yes, a little, not yet'. Both take under five minutes a day and use the same rule in every lesson.
Insight. Two cheap sources that look at behaviour and understanding make a stronger picture than either alone.
The survey that asked for a favour
Scene. Dilnoza in Samarkand writes a survey question: 'Don't you agree that my new homework routine is better?' Her colleague Malika reads it and points out that it asks for a 'yes'. Dilnoza rewrites it as 'Compared with last month, doing homework is: easier / the same / harder', and removes the line for the pupil's name.
Insight. Neutral wording and anonymity let pupils tell the truth instead of what the teacher hopes to hear.
Six notebooks by rule
Scene. Aziz in Fergana wants to see if his new writing frame helps Grade 7 pupils write conclusions. Instead of choosing the best notebooks, he looks at every fifth name in the register: six pupils. He reads the same type of task from the week before the frame and the week after, and notes in one line how many conclusions give a reason.
Insight. A rule for choosing the sample prevents the picture from being flattered by the best work.
Common pitfalls
- Relying only on the end-of-term test, which comes too late to guide the cycle and covers too much.
- Writing leading or double survey questions that push pupils towards the answer you want.
- Collecting the baseline one way and the later evidence another way, so the two cannot be compared.
- Collecting more than you can read: a pile of slips that nobody analyses is wasted time for you and for pupils.
Build an Evidence Plan
- Open the Activity Kit: a SAMPLE question from Grade 5 maths in Bukhara, six evidence cards (A to F), a weak survey and an evidence-plan table.
- Pairs (10 min): rate each evidence card for the sample question: does it fit, is it quick, is it fair, is it kind? Choose the best two or three.
- Pairs (10 min): find what is wrong with each of the four survey questions and rewrite them. Then write one exit-ticket question.
- Groups (10 min): fill the evidence-plan table for the sample question: sources, when, how long, and the counting rule. Compare with another pair.
- Alone (5 min): start the evidence plan for your own question from Session 1 and write your classroom action task from the Kit.
Your facilitator will guide you through this activity in the session.
Questions to discuss
- Which source was cheapest, and which one told you the most? Were they the same?
- Why is a leading survey question a problem for you as a researcher?
- What counting rule did you write, and could a colleague apply it the same way?
Reflect
- What evidence do you already collect by habit but never use for a question?
- Which of your pupils might give a survey answer to please you rather than tell the truth?
- How many minutes a day can you really give to collecting evidence?
Try it tomorrow
- Choose the one question for which a tally would be enough, and write its counting rule in one sentence.
- Write one exit-ticket question for tomorrow and keep it neutral and short.
- Check one survey question you have ever used for a leading or double part.
Further reading and links
Use two cheap, fair sources that fit your question, collect them the same way every time, and keep a one-line log.