Trust, but Verify the P200: A Student’s One-Session Reproducibility Audit

Reproducibility problems rarely announce themselves with a retraction notice. More often they look like a protocol that says standard conditions, a pipette with a faded sticker, or a reagent calculation that is just a little too neat. For UK life-science students and early-career researchers, spotting those red flags early is a practical skill, not a character test.

Why one lab session is enough to start

The UK Reproducibility Network and many university research-integrity teams push the same message: reproducibility is everyday lab craft. You do not need a meta-analysis to improve it. You need a timeboxed audit that fits between an induction and a supervision meeting.

The goal is not to catch anyone out. It is to stop a small ambiguity becoming six weeks of wasted work. Think of it as a pre-flight check for a project, not a performance review.

Red flag: the protocol that says standard

A protocol that says incubate until colour change, centrifuge briefly or use standard buffer is not a protocol. It is a memory test. If two people can read the same method and do different things, the method needs detail.

Look for timings, temperatures, centrifugation speeds, reagent lot numbers, sample volumes and the exact version of the document. A good test is simple: could a new master’s student follow this without asking three questions? If not, write the answers down.

Red flag: kit with no calibration sticker

Pipettes, balances, thermocyclers and plate readers all drift. A missing or out-of-date calibration sticker is a quiet reproducibility risk. The P200 that everyone trusts may be the one nobody has checked since the last lab move.

You do not need to recalibrate everything yourself. Check the sticker, check the log, and flag anything overdue. If your lab runs a quick gravimetric check for pipettes, ask to watch. If not, ask why. That question alone can improve a project.

Red flag: controls that are decorative

Controls are not garnish. A no-template control, a positive control, a vehicle control and biological replicates all answer different questions. If nobody can explain what would make a control fail, it may be there for decoration.

Before you run the experiment, sketch the controls on one page. Write what each one rules out. If a control cannot change your interpretation, it may not be doing much. If it can, you have just built a better experiment.

Red flag: reagent maths that looks too tidy

Suspiciously neat reagent maths is a classic. Stock concentrations, unit conversions, dilution series and final volumes are easy to fumble, especially when a spreadsheet has been copied forward for years.

Recalculate the key steps independently. Use a calculator, not the lab spreadsheet, and check the units out loud. Millimolar versus micromolar is not a personality flaw, but it can ruin a week. Label stocks with concentration, solvent, date and initials. Future you will be grateful.

The one-session audit

Set a timer for 75 minutes. Work through this in order.

• 0 to 10 minutes: read the protocol and highlight every vague phrase. Do not fix anything yet.
• 10 to 25 minutes: check calibration stickers and logs for the pipettes and instruments you will use.
• 25 to 40 minutes: recompute the reagent maths independently, including one dilution series.
• 40 to 55 minutes: map the controls and write what each one rules out.
• 55 to 70 minutes: photograph the bench setup and note any deviations from the protocol.
• 70 to 75 minutes: write a short note to your supervisor or lab manager with three fixes and one question.

That is it. No heroic overhaul. Just a small audit that makes the next experiment easier to trust.

Keep it normal

Reproducibility is not a policing job. It is a lab habit, like labelling a tube or checking a freezer alarm. UK doctoral training programmes increasingly include research-integrity and reproducibility content, but the real work happens at the bench.

So trust your lab, but verify the P200. Check the protocol, the calibration sticker, the controls and the sums. A short audit now is cheaper than a long argument later.