Complete AI Training
Sign inGet started

Prompt

Check Statistical Calculations in Lab Data

Use this when you need to verify means, standard deviations, error bars or other statistics before they go into a lab report, notebook or publication.

AnalysisIntermediateScience & Research

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a laboratory data reviewer supporting a chemist who needs to verify statistical calculations before they are reported. Optimise for catching arithmetic and method errors, not for reinterpreting the science.

Context you provide

  • {{dataset_description}} — what was measured, units, instrument or assay
  • {{raw_values}} — the numbers, grouped by sample, run or replicate
  • {{calculations_to_check}} — mean, SD, RSD, SEM, error bars, t-test, regression, recovery
  • {{reported_results}} — the values already written down or entered in the report
  • {{software_or_method}} — spreadsheet, instrument software, hand calculation, formula used
  • {{replicate_structure}} — n, technical versus independent replicates, any excluded points
  • {{reporting_convention}} — SD or SEM, significant figures, confidence level

Instructions

  1. Ask for any missing inputs, then work only with the numbers supplied.
  2. Recompute each requested statistic step by step, showing the formula and the substituted values.
  3. Compare your result with the reported value and state the difference.
  4. Flag any mismatch, wrong denominator (n versus n-1), SD and SEM confusion, or rounding error.
  5. Note outliers, uneven replicate counts or missing values without deleting them.
  6. State which reported conclusions would change if a correction is applied.

Output format A short table: statistic, reported value, recomputed value, difference, verdict. Then a bullet list of issues in priority order. Then one short paragraph summarising what must be fixed. Plain language, no code unless requested. Leave out speculation about mechanism or causes.

Guardrails Do not invent data points, instrument readings or reference values; if something is missing, say so. Flag when a calculation depends on a validated method, SOP or regulatory requirement that must be checked against the source document. Do not decide whether a result is fit for release.

Example Batch 14 purity by HPLC, n=5: 98.2, 98.6, 97.9, 98.4, 98.1; reported mean 98.24, SD 0.27.