Bring the course together: plan a comparison, inspect the data, and write a conclusion someone else can check.
Field diagramA repeatable study cycle
Estimated reading
8 min
Estimated practice
45 min
Equipment
Supplied data or water-measurement equipment
Progress stays in this page session. Allow lesson storage in Cookie settings to remember it; browser settings may also prevent saving.
Objective
By the end, you can:
Define a comparison before looking at its outcome.
Build a traceable record from raw readings to a conclusion.
Distinguish evidence about repeatability from evidence about accuracy.
The investigation
Imagine comparing two ways of delivering a nominal 10 mL of water. Method A uses a graduated cylinder; method B uses a suitable graduated pipette with a mechanical controller. Each fresh delivery is weighed in a separately tared receiving vessel on the same balance. Which method produces less variable delivered masses in this setup?
Start with the supplied synthetic dataset. It is invented for learning, not a performance specification or a test of a brand. It represents one operator, one session, and six fresh deliveries per method. The sequence alternates method order within pairs to reduce a simple order effect; it does not remove every confounder. Six is a convenient teaching sample, not a recommended sample size for your own research.
Write the plan before the answer
State your outcome as delivered mass in grams, with within-method spread as the comparison. Plan to plot every reading against trial order, then compare means and ranges; sample standard deviations are an optional extension. Preserve the original file and do analysis on a copy. Name the file, method version, and any exclusions before calculating.
Repeated deliveries assess repeatability under these conditions. They do not represent six different operators or six different instruments. A narrower spread does not establish accuracy: neither a validated target mass nor a complete uncertainty assessment is supplied. Do not assume that 10 mL of water always has a mass of exactly 10 g.
Choose your route
Data route: download the supplied file from the field kit, inspect the units and teaching-data labels, and complete the worksheet. All course outcomes can be practiced without laboratory equipment.
Bench route: use only room-temperature water after completing the readiness, records, meniscus, liquid-transfer, and balance lessons. Select two devices suitable for the same nominal volume and a balance able to resolve useful differences. Follow the actual equipment instructions. Write your own sample-size rationale and allocation sequence, label each fresh delivery, and record conditions. If the balance cannot distinguish the differences, report that limitation instead of claiming the devices are equivalent.
What to hand to another reader
Prepare a short report with your question, prediction, equipment or dataset provenance, method, untouched raw data, calculations, plots, conclusion, and limitations. Include unsuccessful trials and deviations with their disposition. Distinguish the plan you made beforehand from questions you noticed afterward. Finish with one change that would make a repeat more informative.
Planning: name the outcome, comparison, experimental unit, and likely confounders.
Technique and records: record enough detail to reconstruct each observation.
Analysis: label axes and units, show all observations, and make calculations traceable.
Interpretation: state what the data support and what remains unknown.
Repeatability: give another reader the files, method version, and proposed next test.
Self-review rubric
Score each of the five areas above from 0 to 3. Zero: absent. One: present but ambiguous or incomplete. Two: clear and traceable, with small gaps. Three: another reader can check the reasoning and repeat the analysis. Rework any area below two. This is a learning rubric, not certification of laboratory competence. A negative or inconclusive finding can earn full marks.
Reveal worked answer: Worked answer · synthetic teaching data
Worked example
Worked answer · synthetic teaching data
Method B has a smaller observed spread of delivered masses in this dataset. The result supports a limited statement about this setup, not an equipment certification. Extra digits below are calculation check values, not a claim of measurement accuracy.
01A: 9.80, 10.04, 9.94, 10.10, 9.87, 10.01 g. Mean = 9.960 g; range = 0.30 g.
02B: 9.98, 9.99, 9.97, 10.00, 9.99, 9.98 g. Mean = 9.985 g; range = 0.03 g.
03Optional sample standard deviations: A ≈ 0.112 g; B ≈ 0.0105 g.
04Conclusion: B produced less variable delivered masses in the supplied six-delivery comparison. No accuracy reference was provided.
05Next investigation: test additional operators and devices with a planned allocation, and specify a suitable reference if accuracy is the question.
Materials
Set out what you need.
Supplied teaching dataset and the field-kit experiment/report worksheets
Notebook or text editor and a calculator or spreadsheet
Optional bench route: room-temperature water, suitable volume devices, mechanical pipette controller, receiving vessels, and a suitable balance
Safety & stop conditions
Pause if the work no longer fits the plan.
For the data route, no laboratory handling is needed.
For bench work, inspect glassware, stabilize the work surface, keep water away from electrical equipment, and have a plan for spills and broken glass. Never mouth pipette.
Use clean water only. Handle and dispose of laboratory water according to the actual workspace rules; do not assume used laboratory vessels are suitable for food or drink. Stop if equipment is damaged or instructions are unavailable.
Method
Work through the steps.
01
Record a question and prediction
Specify the two methods, nominal volume, measured outcome, and what less variation would mean. Keep the prediction even if it is wrong.
02
Preserve the evidence
Keep the supplied CSV unchanged or record fresh observations at the time of measurement. Each row should represent one new delivery, with its trial order, method, units, and any deviation.
03
Plot before summarizing
Put trial order on the horizontal axis and delivered mass in grams on the vertical axis. Use both shapes and labels to distinguish methods. Look for order effects or unusual readings.
04
Calculate and check
Compute the mean and range for each method. Verify one calculation by hand. If calculating sample standard deviation, divide the sum of squared deviations by n − 1 before taking the square root.
05
Report and propose a repeat
State the pattern in this dataset, its limitations, and which further observation would test an alternative explanation. Let another person audit your raw data and calculations.
Checkpoints
Observe, record, investigate.
There are 12 rows: six fresh deliveries per method, not six readings of one delivery.
Every mass is in grams. No conversion from mass to volume is required.
A reference for accuracy is absent; limit the conclusion to the variability observed in this teaching dataset.
Common mistakes
Correct the process, preserve the record.
Calling method B accurate because its values cluster
Clustering describes precision under the stated conditions. Accuracy requires a suitable reference and uncertainty evaluation.
Concluding that all pipettes outperform all cylinders
One operator and one setup cannot establish a population-wide ranking. Repeat with independently selected equipment and operators for a broader question.
Replacing an inconvenient reading
Preserve the original, investigate any recorded problem, and document the basis for exclusion. If the decision was made afterward, disclose it.
Practice
Build your reproducibility package
Use the field kit to make a one-page plan, a results figure, and a short report. Try the calculations before opening the worked example below.
Which differences did the comparison control, and which remain possible explanations?
Can a reader reproduce your summary from the untouched raw file?
What new reference or measurement would be needed to assess accuracy?
How would you extend the design to different operators without pretending repeated readings are independent operators?
Answer: A fresh delivery for the narrow within-session comparison. The same operator and device are reused, so those readings are not independent replicates of operators or instruments.
02Why not convert every mass to milliliters by assuming 1 g/mL?
Answer: The question can be answered in mass units. A volume conversion would need a justified water-density value and conditions, with relevant measurement uncertainty; those are not provided.
03A prediction favored A. Does the B result mean the investigation failed?
Answer: No. A useful investigation preserves and reports evidence that challenges a prediction. The quality of the process matters more than obtaining a preferred result.
