SDAS Student Data Analysis: Eight School Decisions Beyond the Mark Sheet
A mark sheet tells you who scored what; a Student Data Analysis System answers why, what the trend is and what to do next. Eight questions show how standard scores, percentiles, progress rates, multi-year tracking and teacher analysis are used — and how AI natural-language queries let every teacher ask.
On Friday morning the principal and three panel heads cram into a meeting room with the mid-term mark sheet still warm from the printer. The Maths head says, "The mean fell two marks, but I'm sure the paper was harder." The English head scrolls Excel: "I can't see which P5 pupils lag across subjects for two exams in a row." A class teacher raises a hand: "What about the ten middle pupils in my class — does anyone track them?" The principal opens SDAS, asks one question in Chinese, and eight answer sets appear — not another table, but eight issues the meeting can discuss immediately.
A mark sheet records who scored what. SDAS turns scores into school decisions using standard scores, percentiles and multi-year tracking.
Where the mark sheet stops
Most schools have Excel after every test: class means, grade ranks, pass rates. Useful — but four question types stay unanswered: cross-subject comparison (papers differ in difficulty), multi-year tracking (the same pupil from P4 to P6), strand analysis (reading vs writing, number vs geometry) and teacher level (which practices show up in progress). After CSV or Excel import mapped to pupil ID, class, subject and strand, SDAS fills those gaps.
Reading standard scores in two minutes
Standard score (z-score) = (raw mark − grade mean) ÷ standard deviation. 0 is the grade average; +1 is one SD above; −1 is one SD below. Percentile is the share of the grade that pupil scored above. Progress rate is change in standard score between exams, not raw mark difference — paper difficulty shifts make raw deltas misleading. Strand splits each paper by curriculum strand, e.g. "reading comprehension" or "number and algebra".
A P5 pupil's Maths raw mark rises from 68 to 72, which looks like progress. If the grade mean climbs from 60 to 67, the z-score may actually fall slightly — SDAS surfaces that, so panel meetings do not argue past each other about difficulty versus teaching.
Eight management questions
Instead of staring at the whole mark sheet, principals and panel heads can chase eight lines:
Persistent cross-subject lag — pupils with z < −1 in multiple subjects, two or more times in a row; list to class teachers and learning support. Weakest grade strand — lowest mean standard score by strand; sets next term's panel CPD focus. Class progress — which class gained most in z between terms; invite that teacher to share, not to shame others. Multi-year trend — same cohort P4 to P6: who is closing gaps, who is widening them. Pass-rate truth — did the pass rate rise because the paper was easier or because pupils improved? Compare raw and standard scores. Teacher × strand — whose classes improved markedly in which strand; for professional sharing, not one-off league tables. Middle band — pupils at the 40th–60th percentile: is their progress invisible? Remediation impact — tagged list before/after standard-score comparison.
The value of data is not seeing a score but having eight answers you can discuss in the same meeting.
Three steps from Excel to query
Import — upload CSV or Excel; map pupil ID, class, subject, strand. Compute — standard scores, percentiles, progress rates; absences and exemptions by rule. Query — default dashboards plus natural language. Panel heads need not memorise formulas; in a panel meeting ask live, "Which P5 English classes gained more than 0.3 SD in reading?" Project the chart and start the conversation.
Starting professional dialogue with data
Picture the English panel meeting after mid-terms: instead of debating whether "the mean dropped because the paper was harder", the head projects a strand chart — reading comprehension z-scores by class — and asks the teacher whose line climbed 0.4 SD what changed in January. That is the culture SDAS enables when numbers start conversations rather than close them.
Three governance habits keep SDAS from becoming an accountability weapon: trend before single points — one exam fluctuation means little; look for patterns across two or three sittings. Data starts dialogue, it is not the verdict — if one class improved in "number and algebra", the next sentence is "what did you do?" not "why didn't the others". Scoped access — principals see the school, panel heads their subject, class teachers their class; privacy requirement that also keeps talk actionable.
Before-and-after for grant reporting
Schools reporting AI education grants often need before/after evidence for "student learning outcomes". In Edor.ai, SDAS dashboards can sit beside AI usage logs so a panel head can show not only that pupils used a tool, but whether standard scores in the targeted strand moved for participating classes versus comparison groups — one export for the governor folder instead of three incompatible spreadsheets. SDAS multi-year tracking exports standard-score change before and after AI use, participating versus comparison classes, strand-level progress rates. Panel heads spend time explaining why something worked, not merging pupil IDs in Excel.
Common misreadings
-
The claim: SDAS replaces the mark sheet. What actually happens: Import still starts from mark sheets; SDAS is the analysis layer, not the assessment. Inconsistent import formats break calculations.
-
The claim: Standard scores label pupils permanently. What actually happens: Z-scores show relative position for trends and support decisions; conversation should focus on next steps, not public league tables. Access control is non-negotiable.
-
The claim: Natural-language queries replace the data lead. What actually happens: Queries lower the barrier, but import rules, strand mapping and anomaly interpretation still need one or two colleagues who maintain the system — SDAS lets more people ask, not nobody manage.
Three things to try on Monday
- Import your latest exam results as CSV into SDAS and verify pupil ID, class, subject and strand columns map correctly.
- Before the panel meeting, pick three of the eight questions above and run one natural-language query for each; screenshot for projection.
- Agree with the principal how teacher analysis will be used: professional sharing and support, not one-off ranking or blame.
Questions that keep coming up
For each exam it computes standard scores, percentiles and progress rates, grouped by grade, class, subject, strand and teacher. Absences and exemptions follow school rules so staff do not rebuild Excel each time.
Yes. Natural-language queries let teachers ask in plain language, e.g. "Which P5 pupils dropped more than 0.5 SD in Maths between terms?" The system builds the query, shows charts and adds a short explanation.
That depends on school culture. A healthy use finds classes that improved markedly in specific strands and invites those teachers to share practice — data opens professional dialogue, not automatic blame.
Related notes
Reclaiming 8–10 Hours a Week: AI Admin Agents for Notices, Minutes and Parent Letters
Hong Kong teachers spend about 8–10 hours a week on paperwork. Six eAdmin scenarios show how AI drafts eNotice, eCircular, parent letters, meeting minutes, event proposals and iPortfolio comments — and why grounding in the school knowledge base and matching the school's voice decide whether a first draft is usable.
Keep reading
The PIE Cycle Meets AI Dashboards: Evidence for School Planning, Implementation and Review
Every school runs Plan–Implement–Evaluate, but evaluation often relies on impressions. See how PIE KPIs link directly to platform data — AI usage, objective mastery, teacher training records, SDAS results — so planning has a baseline, implementation has tracking, evaluation has evidence, and grant interim and final reports export from the same cycle.
Keep reading
The School Knowledge Base (RAG): AI That Looks Up Your Documents Before Answering
Generic AI has never read your school rules, pacing charts or notice formats. A school knowledge base uses RAG (retrieval-augmented generation) to search school documents before answering and attach source citations. This article covers how it works, what to upload, permission tiers and export requirements, and six practical scenarios for teaching and administration.
Keep reading