About Mizan: how it works, and its limits
Mizan is a short test that adapts to the child. It estimates their level in reading, comprehension and language so we can choose the right practice and books, and follow their growth through the year.
In short: the result is currently a preliminary estimate of level, for practice purposes.
Who it is for
Children from Grade 1 to Grade 9. The question bank now holds more than 3,000 questions across the nine grades. It is not yet designed for «Arabic for non-native speakers» learners; their progress appears in their own track report.
What it measures
- Letters and sounds (Grades 1 and 2)
- Vocabulary and opposites
- Understanding literary texts
- Understanding informational texts
- Language: spelling and grammar
What it does not measure
Speaking, listening to long texts, handwriting, free writing and oral reading fluency. These are followed in the lessons: in the writing and speaking activities, and in reading recordings made with parental consent.
Reference
Every question is linked to a learning outcome from the «General Framework of Arabic Language Curriculum Standards» (the national document), and to the grade where it is taught.
How it works
- It uses the Rasch model from item response theory: each question is chosen based on the previous answer, from the least-covered area, at a difficulty close to the current estimate of the child's level.
- It stops after 30 counted questions, or after 20 if the estimate has settled. There is no time limit.
- An answer given too fast to have read the question is not counted, so up to 40 questions may be shown. If rapid answers reach 30%, the result is marked «unreliable» and does not change the child's level.
Score and reading level
The Adabi score is one scale running from Grade 1 to Grade 9, so a child's scores can be compared with each other across years. The reading level (from 3 to 29) is derived from the score and decides which books the library suggests.
Question difficulty and calibration
Each question's difficulty is currently a preliminary estimate based on its grade and outcome; no calibration study on a large sample of students has been carried out yet. We collect the share of correct answers for every question to calibrate its difficulty over time, and will report the calibration results here when they are complete.
That is why we show no accuracy or confidence figures, and the «expected for grade» range in the report is also provisional.
Limits of use
Use it to choose practice and books and to follow growth. On its own it is not suitable for diagnosing learning difficulties, or for pass/fail or admission decisions.