Switch course ▾
Session 1. School Self-Evaluation Against a Framework
Objective. Rate a school against quality indicators with level descriptors, gather three kinds of evidence for each judgement, moderate ratings with a critical friend and write a short, honest self-evaluation statement.
A self-evaluation is only as good as the way you read the level descriptors and the evidence behind them. Most schools do not fail because they lack a framework; they fail because words like 'most' and 'some' mean different things to different people, and because the evidence they gather is the evidence that is easiest to find. This lesson gives you ways to read descriptors the same way as your colleagues and to notice the biases that make a school look better than it is.
Key ideas
Why a framework
- A quality framework lists the areas of school life that matter and describes what each level of quality looks like. The school can then judge itself against the same words every year, not against feelings. Real examples are How Good Is Our School? 4 (Scotland) and Looking at Our School 2022 (Ireland).
- Course 12.2 (Data-Informed School Improvement) taught how to read data. Here data is one input to a judgement about whole areas: teaching and learning, attendance and wellbeing, leadership, partnership with families.
- Adapt, do not copy. Use the national curriculum framework and ministry regulations as your anchor, borrow the structure of a foreign framework, and choose four to six indicators for the year. A full framework is too big for one cycle.
From evidence to a level
- For each indicator gather three kinds of evidence: what we see (lesson visits, pupils' notebooks), what we measure (marks, attendance, test summaries) and what people say (pupils, staff, parents). One kind alone can mislead.
- Choose the level by best fit: read the descriptors from the top and ask which one matches most of the evidence. The weakest strong evidence stops you from rating higher; one lovely lesson does not make a level.
- Write the reason in one sentence and name the evidence, for example 'Level 3 because 9 of 12 visited lessons had a clear aim and 22 of 30 notebooks had next-step comments.' A rating without a reason cannot be checked.
Moderation and challenge
- Leaders tend to rate their own areas too generously. Ask a critical friend, such as a colleague from another subject or a neighbouring school, to read the evidence and give an independent rating before you compare.
- Use three challenge questions: What is your evidence? What would change your mind? Who would disagree, and what would they say? Challenge the evidence, never the person.
- If two raters differ by one level, look at which evidence each used. If they differ by two levels, they almost certainly looked at different evidence or read the descriptors differently; settle that before arguing about the level.
The self-evaluation statement
- For each indicator write a short statement: the level, two or three strengths, one or two areas to improve, the evidence used and what you do not yet know. One page per indicator is enough.
- Self-evaluation is for improvement. If a school rates itself high to please the department, the plan will fix nothing. An honest Level 2 with a clear next step is more useful than an inflated Level 4.
- Keep paperwork light and reuse it: the same statements feed the improvement plan (session 2) and the annual report (session 5). Respect teachers' time and keep names of pupils and colleagues out of written statements.
Three kinds of evidence for one judgement
Seen
What we observe
- Lesson visits
- Pupils' notebooks
- Routines in corridors
Measured
What we count
- Attendance by weekday
- Term marks
- Test summaries
Said
What people tell us
- Pupil questionnaire
- Staff survey
- Parent comments
Making 'most' and 'some' mean the same thing
Level descriptors use quantity words: almost all, most, many, some, a few. Two people reading 'most lessons have a clear aim' can mean 55 percent or 85 percent. Before rating, agree locally what each word means, for example 'most' is at least 8 of 12 lessons visited and 'almost all' is at least 11 of 12. Write the agreement at the top of the rating sheet.
Then read the descriptor from the top level down and stop at the first phrase that the evidence does not meet. Descriptors often combine conditions with 'and': pupils get feedback and use it. If only the first half is true, the level is not reached. Marking the words you could not prove keeps the rating honest and shows exactly what to improve.
Biases that inflate a rating
The halo effect lets one strong lesson colour the whole picture. Recency lets the last visit weigh more than the first eleven. Self-selected evidence, such as a parents' meeting where the interested families come, looks like a sample but is not. A fourth bias is institutional: leaders who wrote the plan find it hard to rate it low.
The remedies are simple and cheap. Decide the sample before looking, for example every third lesson on the timetable. Use two raters and compare. State what each piece of evidence cannot tell you. Bring in a critical friend whose own area is not being judged. The OECD working paper on school evaluation is a useful reminder that evaluation systems differ widely, so build trust in your own process before you rely on it.
Examples from the classroom
What does 'most' mean in Andijan?
Scene. A deputy head in Andijan and two subject leads rate 'most lessons have a clear aim' from the same twelve visits. One says Level 3 because 7 lessons had a clear aim. Another says Level 2 because 'most' means at least 9. They stop and agree: most is 8 of 12, almost all is 11 of 12. With that rule 7 of 12 is Level 2, and they write 'one more lesson needed to reach Level 3' on the sheet.
Insight. Agreeing numbers for quantity words removes arguments about the level and shows how close the school is to the next one.
The last visit in Namangan
Scene. In Namangan, Dilnoza rates teaching from lesson visits. Eleven visits in September and October were ordinary, but the last one, in November, was excellent and is the one she remembers. She tells her critical friend, Bobur, 'I see strong teaching.' Bobur asks to see all twelve notes. The tally shows 6 of 12 with a clear aim. Dilnoza revises her rating from 4 to 2-3 and writes the sample into the statement.
Insight. Recency makes the last observation weigh too much; looking at the whole record corrects it.
Swapping files between two Samarkand schools
Scene. Two Samarkand schools swap anonymous evidence files for attendance and wellbeing, with school names removed. Each head rates the other school's file and compares with the owner's rating. One head sees that the other school rated Level 4 even though 4 of 11 persistently absent pupils had not been contacted. The owner had not counted the uncontacted pupils as evidence. The heads agree to add that count to every file next term.
Insight. An outsider reading the same evidence quickly spots the evidence the owner overlooked.
Common pitfalls
- Using words like 'most' without agreeing what they mean, then arguing about levels.
- Rating from the evidence that is easiest to collect, such as the best lessons or the parents who come to meetings.
- Treating the critical friend's challenge as an attack and defending instead of listening.
- Copying a foreign framework in full, so that the school spends the year on paperwork.
Rate, Moderate and Write the Statement
- Trios (5 min): open the Activity Kit. You have a SAMPLE four-level framework with two quality indicators (Table A) and an evidence file for a sample school of 520 pupils (Table B). Agree who is the leader, who is the critical friend and who keeps time. Roles rotate in the second round.
- Alone, then trio (15 min): each person silently rates both indicators using the rating sheet (Table C): level, evidence numbers used and one sentence of reason. Do not talk yet. Then compare. Where you differ, say which evidence you used and what descriptor you matched.
- Trio (15 min): agree one rating per indicator. Mark each evidence item strong, medium or weak (is it a count, a sample, a single comment?). Decide which indicator rests on thinner evidence and what one extra piece of evidence would settle it.
- Challenge round (10 min): two trios swap sheets. The visitors ask the three challenge questions (What is your evidence? What would change your mind? Who would disagree?) for 5 minutes, then the hosts do the same for the visitors. Change a rating only if the evidence supports it.
- Part B (15 min), alone: write the self-evaluation statement for ONE indicator on the statement form: level, two strengths, one area to improve, evidence used, what we do not yet know. Check that it names evidence, not people, and that the weakness is stated plainly.
- Whole group (10 min): two people read their statements aloud. Others note one phrase that is evidence-based and one that is only an impression, and suggest how to fix it. Check the sample key last.
- Alone (15 min): write your leadership task from the Kit: which indicator of your own school you will self-evaluate, which three kinds of evidence you will collect, who will be your critical friend and the date you will bring back the statement.
Your facilitator will guide you through this activity in the session.
Questions to discuss
- Where did you and a colleague differ most, and was it the evidence or the reading of the descriptor?
- What was it like to be challenged on your rating, and how would you make that feel safe in your school?
- Which part of your own school do you suspect you rate too kindly, and who could look with fresh eyes?
Reflect
- Which quality indicator would your teachers say matters most, and which would your pupils?
- What evidence does your school never collect about a very important area?
- Who would give you honest challenge, and what would make it easy for them to do so?
Try it tomorrow
- Write numbers for 'almost all', 'most', 'some' on one line at the top of your next rating sheet and agree them with a colleague.
- Choose your sample of lessons or notebooks before you look, for example every third notebook in a pile, and note how many you took.
- Ask one colleague from another subject to rate one indicator from your evidence, then compare differences on a single page.
Further reading and links
Rate by best fit against level descriptors, name three kinds of evidence for every judgement, invite challenge from a critical friend and write the weakness down plainly.