Safety and evidence boundary
Use only teacher- or parent-approved low-risk investigations, supplied datasets, paper models, or simulations. No human-subject experimentation, personal health data, biological culturing, unknown environmental samples, hazardous chemicals, animal manipulation, ingestion studies, or medical claims are permitted. If a proposed investigation cannot meet these limits, convert it to a data-analysis or model-based study.
Explicit science instruction
A trustworthy investigation preserves the distinction between raw data, cleaned or transformed data, and interpretation. Record units, missing values, unusual observations, procedural deviations, and any exclusion rule. Never delete inconvenient data merely because they weaken the expected pattern.
Replication tests whether a result is consistent under repeated conditions. Reproducibility asks whether the documented method is clear enough for another person to repeat or independently analyze. A method can improve during a study, but revisions should be dated and disclosed.
Vocabulary: raw data; transformed data; data provenance; exclusion rule; replicate; reproducibility; anomaly; protocol deviation; revision log.
Worked example with reasoning
A student records five replicate measurements: 8.1, 8.4, 8.2, 14.9, 8.3. Instead of deleting 14.9, the student checks the record. If a documented instrument error occurred, the value may be excluded under a stated rule while preserving it in the raw table. If no error is known, it remains part of the evidence and is discussed as an outlier.
Guided question
Why is preserving a questionable measurement often more scientifically responsible than silently removing it?
Independent work
- Complete or analyze the approved investigation/data set.
- Preserve an untouched raw-data table.
- Create a second analysis table if transformations are needed.
- Record every missing value, anomaly, or method deviation.
- Write a dated revision note.
- State whether the method is reproducible from your documentation.
Misconception check
“Cleaning data” does not mean removing values that disagree with a hypothesis. Any exclusion requires a defensible rule tied to measurement quality or study design, not the desired conclusion.
Formative understanding check
What information belongs in a revision log?
Show the worked response
What changed, when it changed, why it changed, what evidence prompted the change, and whether the change affected data already collected.
Open-book formative practice only; not an official mastery attempt or grade.
Mentor evidence conference
Ask Archaemenes to challenge the alignment among your question, method, data, graph, claim, and limitations. The mentor may critique reasoning but does not approve unsafe procedures or provide protected assessment answers.
Open Archaemenes for this lessonTeacher guidance
Foundation: use a supplied raw table and mark anomalies without deleting them. Core: maintain raw versus analysis tables and document all method revisions. Extended: write a preregistered-style exclusion rule and evaluate whether the final analysis followed it.
Preserve learning evidence
Keep the research question, model, method, data display, CER, correction record, and limitations in the matching Week 17 evidence field and semester portfolio.
Return to thursday Evidence Workspace