Student Confidence

How confidence is calculated

For every grammar tag a student has touched we keep two weighted counts: Right (evidence from correct answers) and Wrong (evidence from incorrect answers). The score is just Right / (Right + Wrong) × 100.

Score (0 – 100)

Starts at 50 for every new tag — a neutral "no evidence yet" position. It moves up or down as the student answers questions:

• Above 50 — more right than wrong, weighted by question difficulty and source reliability.
• Below 50 — more wrong than right.
• The weighted nature means one bad exam day can't instantly crash a solid score, and one lucky guess can't launch a weak one.

Status bands

The score is grouped into six bands so you can scan a student's profile quickly:

• not competent 0 – 25 — clear gap, no evidence of competence
• flagged 26 – 39 — needs immediate attention
• struggling 40 – 54 — having difficulties
• progressing 55 – 69 — making progress
• confident 70 – 84 — good understanding
• competent 85 – 100 — mastered

Certainty

How much data backs the score. Attempts = Right + Wrong (weighted).

• very low fewer than 5 weighted attempts — essentially the starting position, treat with caution
• low 5 – 9 attempts — a few interactions
• moderate 10 – 19 attempts — solid practice
• high 20+ attempts — extensive practice

How the score moves

After each answer the system calculates what was expected given the student's current ability, the question's difficulty, and the number of answer options (so guessing on a 4-option multiple choice doesn't earn full credit). The change applied to the score is larger when the outcome is surprising — a hard question answered correctly, or an easy question missed — and smaller when it's expected. This is the Bayesian-Elo method: the same idea used in chess ratings, adapted for language learning.

Source weights

Not every answer carries the same weight. Teacher-graded work counts for more than a quick tap in a gamified exercise:

• Writing & speaking — strongest weight. A teacher has judged the response directly.
• Exams — strong. Timed and standardized.
• LMS homework — moderate.
• Learning path — small per tap, but many quick interactions add up over time.

In practice, one writing submission moves the score roughly three times as much as one learning-path tap.

Trend

Compares the score over the student's last five recorded updates: improving, stable, or declining.