The IB Biology IA Data Collection Checklist

Most Data Analysis marks are lost before data collection even begins — in weak planning. This checklist covers the 30 items you should verify in the last 72 hours before you start the main experiment.

Work through it in order. If any item is unchecked, pause and fix it — every unchecked item typically costs 0.5–1 mark on the 24-mark IA.

---

A. Research Design Foundation (pre-experiment)

Check these against the Research Design criterion (6 marks):

• Research question is focused and names the independent variable, dependent variable, and organism/system explicitly.
• At least 5 levels of the independent variable (not 3, not 4 — the minimum is 5 for trend analysis).
• Variable levels are justified — why these 5 specific values? Why evenly spaced? Why this range?
• All control variables listed with tolerances (e.g. temperature 25 ± 0.5 °C, not just "room temperature").
• Statistical test pre-registered in Research Design (see our statistical tests guide).
• Biological rationale for the predicted trend — what mechanism makes you expect this result?
• Ethics/safety section completed — CLEAPSS / school ethics / organism welfare documented.

B. Replication and Sample Size (the 5×5 minimum)

• At least 5 replicates per level of the independent variable (the "magic 25" floor).
• Replicates are genuinely independent — separate organisms, separate beakers, separate trials (not re-reading the same sample).
• Randomised order of measurement to avoid systematic drift (e.g. if you measure all "low" samples first, they will be colder or drier than "high" samples).
• Calibration standards run at the start and end (for pH meters, spectrophotometers, balances).

For categorical experiments (e.g. PTC tasting), substitute: at least 2 categories with ≥10 replicates each.

C. Uncertainty Budget

• Every instrument has a documented uncertainty (e.g. ± half the smallest division on a ruler, ± 0.01 g on a 3-decimal balance).
• Uncertainty is propagated through calculations (absolute for addition/subtraction, percentage for multiplication/division).
• Biological variability is recorded — standard deviation per group, not just a single mean.
• Worked calculation example in the appendix for each derived quantity.

D. Measurement Protocol

• All raw data recorded with units (not "2.3" but "2.3 cm³ s⁻¹").
• Timestamps recorded for time-sensitive measurements (enzyme reactions, respiration, plant growth).
• Ambient conditions logged once per session — room temperature, humidity, barometric pressure (if relevant).
• Measurement repeated if suspicious (don't delete; re-measure and record both).
• Data logger output saved as a separate raw file — Excel "auto-correct" has erased scientific notation more than once in published history.

E. Reproducibility

A peer should be able to repeat your experiment using only your method section. Examiners test this by asking: "Could I replicate this?"

• Specific equipment models and sizes listed (not "beaker" but "250 mL borosilicate beaker").
• Reagent concentrations and suppliers listed where relevant.
• Software version noted for any computational analysis.
• Photographs of the experimental setup in an appendix.
• Method written in impersonal past tense — "10 mL of 0.1 M NaCl was added…"

F. Pilot Trial

Run a pilot 2 weeks before the main experiment with a reduced version (2 levels × 2 replicates).

• Pilot completed and reviewed.
• Technical issues identified and fixed (e.g. "my gas syringe gave negative readings at low rates").
• Timing calibrated — does the full experiment fit into a single lab session, or do you need 2 sessions?
• Method revised based on pilot; old version archived.

G. Data Hygiene During Collection

• Lab notebook or digital record updated in real-time, not reconstructed after.
• Anomalies flagged as they occur (e.g. "sample 3 had air bubble").
• No data deletion — if you exclude a point later, you must justify it in Evaluation.
• Backup of raw data after each session (cloud + local).

H. Post-Collection Sanity Checks

• Descriptive statistics computed — mean, SD, SE per group.
• Quick scatter or box plot reviewed — does the trend make biological sense?
• Outliers identified using a pre-defined rule (e.g. Tukey 1.5 × IQR).
• Data still looks like a realistic biology experiment (not suspiciously clean).

---

What Goes Wrong at Each Stage

---

The 2025 Rubric — Exactly What to Show

The 2025 IB Biology IA rubric looks for specific observable patterns in your report. Here is how this checklist maps to rubric signals:

Research Design 5/6 signal:

"Five levels of independent variable (IV) were chosen to span the range of expected biological response, with equal spacing to enable trend fitting. The control variables in Table 1 were monitored continuously using instruments listed in Table 2, with tolerances derived from manufacturer specifications."

Data Analysis 6/6 signal:

"Raw data (Table 3) include all 30 individual measurements with instrument uncertainties. Processed data (Table 4) include means ± SE, with a worked example of the initial-rate calculation in Appendix A. Normality was verified by Shapiro-Wilk (p = 0.21) and homogeneity of variance by Levene's test (p = 0.48), supporting the parametric one-way ANOVA used in the analysis."

Evaluation 6/6 signal:

"The largest source of systematic error was temperature drift during the 40 °C trials (mean drift 0.8 °C over 30 minutes), which would have reduced measured rates by approximately 4-6% based on a Q₁₀ of 2. A thermostatically controlled water bath with active feedback would reduce this to ±0.1 °C."

---

If You're Already Behind Schedule

If you are less than 2 weeks from submission and have not collected enough data, read the IA troubleshooting guide before re-running experiments. Many fixable problems do not require more data — they require re-framing the research question or statistical approach.

For a full walkthrough of the 2025 rubric criteria, see the IA rubric breakdown.