Sarawak deep dive

Each row is one Pejabat Pendidikan Daerah (PPD). Cohen's d shows how much better SEEP students in that PPD scored compared to a matched control. The tier column turns the number into a plain-language verdict: Excellent (≥1.20), Strong (≥1.00), Large (≥0.80), Below target (<0.80).

3 PPDs are below the Large-effect floor. These districts are getting less benefit per student than the rest of the state. Usually this is driven by teacher turnover, unreliable Module B delivery in rural areas, or low attendance cycles — not by students being less motivated.

Priority list (bottom up): Lundu, Dalat, Julau. Pair each with its strongest geographic neighbour for teacher coaching and re-measure after one cycle.

Schools are placed into one of five attendance buckets based on their mean attendance rate this semester. Each bar is the average gain score (post-test minus pre-test) for schools in that bucket. A steeply rising pattern means attendance drives learning; a flat pattern would mean attendance doesn't matter.

The rise is steep and monotone — every attendance band outperforms the one below it. Attendance is not a nice-to-have; it is the single strongest lever SEEP has. The ~7 schools under 85% are leaving 9+ points of gain on the table per student, which compounds across the 108,469-student cohort to tens of thousands of missed opportunities.

Focus enforcement attention on the 7 schools in the <85% bucket first — they have the largest upside. Pair them with the 28 schools in the 99–100% bucket for playbook-sharing visits.

Grey bars are the risk distribution at the last measurement cycle; green bars are today. If the green On-track bar is higher than the grey one, students are graduating UP the ladder (good). If the green At-risk bar is higher, students are sliding DOWN (bad). The delta tiles below show each movement in plain numbers.

4,885 students have exited the At-risk bucket since last cycle — these are the real SEEP wins, not the headline averages. But 4,309 students moved out of Watch, which combined with the On-track growth means the interventions are graduating students upward rather than just capping the top. This is the healthiest possible flow pattern.

Protect whatever the at-risk intervention team is doing — it's working. Sample 100 students who moved from At-risk → Watch this cycle and document what changed in their attendance, test scores, and Module B completion. That playbook is your scaling asset for the next semester.

Each row is one Tingkatan. The Male (d) and Female (d) columns show Cohen's d separately for each gender. The Gap column is the difference (positive = female higher, negative = male higher). A gap under 0.05 means the SEEP effect is effectively balanced; larger gaps signal an equity concern.

The effect tails off gradually from Tingkatan 1 (d=1.41 for female) to Tingkatan 5 (d=1.19 for male). This is the classic 'earlier intervention is more efficient' pattern — students who enter SEEP in lower Tingkatan have more time and plasticity to benefit. The small but consistent female lead (~0.05 pts) is typical in Malaysian education analytics and is not a red flag. Both halves of the cohort are clearly benefiting.

Protect the Tingkatan 1 and 2 funnel — that's where the highest marginal return is. For the Tingkatan 5 cohort (smallest effect), run a follow-up analysis to see whether gain erosion correlates with exam pressure or attendance drop in the final stretch.