Cognia School Data Analysis

Turn a school's raw data into accreditation-ready analyses following Cognia's Self-Assessment Workbook methodology. The goal is never to present raw data — it is to present analyses (evidence → findings → theories of action).

When to use

• The user provides school data (assessment results, surveys, observation logs, handbooks, plans) and wants analysis, findings, or an accreditation narrative.
• Preparing for a Cognia Accreditation Engagement Review self-assessment.
• Running a continuous-improvement data cycle (status / improvement / growth, subgroup gaps, root-cause analysis).
• Writing the three required analyses or the Executive Summary.

Core principle

Data → (analyze → synthesize) → information → findings (factual, conclusive statements) → evidence (only the info that supports findings) → theories of action (if/then/so that). Never jump to action before root cause. Treat symptoms, not causes.

The five-step process

1. Understand the purpose. Agree to honesty/transparency; commit to using results for improvement.
2. Identify & evaluate data sources. Inventory every source using the four categories below. Use assets/data_source_inventory.md as a worksheet.
3. Clean data. Drop irrelevant sources; verify specificity, validity, accuracy. Disaggregate to specific items/pages, not whole documents. See references/methodology.md (cleaning checklist + common mistakes).
4. Analyze & synthesize. Cluster cleaned data into themes (start with the four Key Characteristics: Culture, Leadership, Engagement, Growth). Tag each point positive/negative.
5. Interpret, prioritize, act. Write finding statements; prioritize by capacity; run root-cause analysis; draft if/then/so-that theories of action.

Four data-source categories (use for the inventory)

• Documentation — handbooks, strategic/action/tech plans, policies.
• Observations — eleot®, teacher observations, meeting evaluations.
• Perceptions — surveys, focus groups, interviews (learners, families, educators).
• Performance — organizational metrics + student academic/non-academic data (attendance, discipline, assessment results).

Required analyses (do all three)

Each follows the same five steps and produces a three-part narrative: (1) evidence analyzed & synthesized, (2) findings (noteworthy achievement + areas for improvement), (3) root-cause interpretation + theory of action.

• Student Performance Analysis — prompts, EC1–EC4 rubric, and output template in references/analysis_prompts.md and assets/student_performance_template.md. Required for schools (not for standalone corps / ESOs / charter authorizers / SEAs).
• Stakeholder Feedback Analysis — must cover learners (grade 3+), families, educators. Prompts + EC1–EC4 in references/analysis_prompts.md and assets/stakeholder_feedback_template.md.
• Learning Environment Observation Analysis — uses eleot® or equivalent that measures learner engagement (NOT a teacher observation tool). Prompts + EC1– EC3 in references/analysis_prompts.md and assets/learning_env_template.md.

Rate each analysis package holistically against its evaluative criteria (detailed rubrics in references/evaluative_criteria.md) to set confidence and flag improvement areas.

Findings, root cause, theories of action

• Finding statement: "Based on the information we analyzed and synthesized, we found that …" — factual, no solution embedded.
• Root cause: fishbone (categories: Curriculum, Instruction, Engagement, Assessment, etc.) or the Five Whys. Data show symptoms, not causes.
• Theory of action: "If [action], then [expected change] so that [outcome]." Provide ≥1 for improvement and ≥1 to sustain a strong practice.

Working with real data files

When the user supplies CSV/XLSX exports, run scripts/analyze_school_data.py to produce a structured brief (status, improvement trend, growth, subgroup gaps) the agent uses to write the narrative. Then map results into the relevant assets/*_template.md and apply the EC ratings from references/evaluative_criteria.md.

python scripts/analyze_school_data.py \
  --file results.csv --year-col Year --group-col Subgroup \
  --value-col Proficiency --category-col Subject \
  --out brief.md

See scripts/analyze_school_data.py --help for all options.

Executive Summary

After the three analyses, synthesize a ~2,000-word institutional narrative using assets/executive_summary_template.md (History, Community, Governance, Mission/Vision/Values, Enrollment, Philosophy, Curriculum, Instruction, Personnel, Student Performance, Improvement Initiatives).

Quality checklist (before delivering)

• [ ] Raw data not presented as the analysis; analyses/synthesis shown.
• [ ] Sources cleaned and categorized (D/O/P/P); irrelevant data dropped.
• [ ] Each analysis rated on its EC rubric; confidence stated.
• [ ] Findings factual, separated from solutions; prioritized by capacity.
• [ ] Root-cause analysis performed (not just symptom treatment).
• [ ] ≥1 improvement + ≥1 sustain theory of action (if/then/so that).
• [ ] Subgroup/achievement-gap disaggregation included where relevant.
• [ ] Narrative follows the three-part outline; evidence cited specifically.

Reference files

• references/methodology.md — full five-step detail, cleaning checklist, common data mistakes, fishbone/five-whys guidance.
• references/evaluative_criteria.md — all EC rubrics (Student Performance EC1– EC4, Stakeholder EC1–EC4, Learning Environment EC1–EC3).
• references/analysis_prompts.md — exact prompts for each analysis's Areas of Noteworthy Achievement / Areas in Need of Improvement.
• assets/data_source_inventory.md — four-category source worksheet.
• assets/student_performance_template.md, assets/stakeholder_feedback_template.md, assets/learning_env_template.md — narrative templates.
• assets/executive_summary_template.md — executive summary outline.