Make measurement uncertainty useful in a Chemistry IA
Work through two original Chemistry IA examples, from comparing concentration bounds to diluting a sample into a calibration range.

Identify the quantity you really measure
Begin with the raw measurements behind your result. A titration volume comes from a difference between readings; a mass change comes from two masses; a concentration may depend on several quantities. List the instrument, unit, reading method and relevant uncertainty information beside each raw value. Follow your teacher's conventions for the apparatus and calculation involved.
Do not assign the same uncertainty to every measurement because the table looks tidier. Instrument resolution, calibration information and variation between repeated observations describe different aspects of uncertainty.
Distinguish variability from a possible bias
Repeated measurements help you inspect variability, but they do not automatically expose every systematic effect. A consistently miscalibrated balance may give closely grouped readings. An endpoint judged differently by different observers may create another pattern. Explain what your actual records suggest instead of labelling all discrepancy as random error.
Keep qualitative observations with the data. A slow endpoint change, drifting sensor or inconsistent preparation can help you explain why scatter appeared and which part of the method needs attention.
Show one propagation calculation clearly
As an illustrative classroom calculation, suppose each burette reading is assigned an uncertainty of ±0.05 mL. Under a conservative addition convention for the difference, the delivered volume has an uncertainty of ±0.10 mL. For 24.60 mL, this corresponds to approximately 0.41 percent relative uncertainty. State the convention used; other justified uncertainty models may combine information differently.
This example shows why the uncertainty of a delivered volume is not simply copied from one reading. Use the instrument information and approach agreed for your actual investigation.
Compare contributions before proposing an improvement
Express suitable contributions on a common relative basis when the calculation requires it. An illustrative 0.80 g measurement with ±0.01 g uncertainty has 1.25 percent relative uncertainty. Measuring a larger quantity with the same absolute uncertainty can reduce the relative contribution, provided the method and apparatus remain appropriate.
Identify the largest meaningful contribution and check whether changing it would improve the final result. Buying a more precise instrument may have little value if sample variation or endpoint judgment dominates. A proposed improvement should address the limitation established by your measurements.
Original simulated measurement case
Worked Chemistry IA example: which measurement should improve?
Two hypothetical solutions contain the same solute. Solution P has a recorded solute mass of 0.500 g; Q has 0.510 g. Both have a recorded final solution volume of 25.00 mL. Their calculated mass concentrations are 20.0 and 20.4 g L−1. Can the measurement model distinguish which is more concentrated?
Assumptions for this arithmetic exercise: each solute mass is assigned a possible error bounded by ±0.001 g. Compare two fictional volume methods below. Allow each solution's inputs to vary separately within those bounds; do not assume cancellation of shared errors. Assume complete dissolution and that the recorded mass represents the solute. These are invented data and bounds, not real apparatus specifications, experimental results or a submitted IA.
Define the quantity: mass concentration = solute mass / final solution volume. Use litres in the denominator: 25.00 mL = 0.02500 L, so P gives 0.500 g / 0.02500 L = 20.0 g L−1. The denominator is final solution volume, not simply the volume of solvent added.
Volume bound: ±0.50 mL
- P: calculated concentration range
- 19.56–20.45 g L−1
- Q: calculated concentration range
- 19.96–20.86 g L−1
Ordering unresolved. Allowed values overlap: Q's calculated central value is higher, but this model also permits Q to be less concentrated than P.
Volume bound: ±0.05 mL
- P: calculated concentration range
- 19.92–20.09 g L−1
- Q: calculated concentration range
- 20.31–20.49 g L−1
Ordering resolved within this model. Q's lowest allowed concentration exceeds P's highest. The narrower volume bound supports the comparison under the stated assumptions.
Displayed lower endpoints are rounded down and upper endpoints up to 0.01 g L−1, so rounding does not narrow the calculated bounds.
Check one range. For positive mass and volume, the smallest ratio uses the smallest mass and largest volume; the largest ratio uses the largest mass and smallest volume. For P with the ±0.50 mL bound:
Why the volume improvement matters. For P, the assumed mass bound is 0.001 / 0.500 × 100 = 0.20%. The first volume bound is 0.50 / 25.00 × 100 = 2.0%; the second is 0.20%. Under this model, improving only the balance would leave the larger volume contribution largely untouched. Once both contributions are similar, the next improvement needs a fresh comparison.
Bounds are not statistical significance. This is interval arithmetic with assumed limits, not a confidence interval or a significance test. It assigns no probability to an ordering. Purity, transfer losses, calibration effects and temperature-dependent volume changes would need appropriate treatment in a real investigation; repeated readings alone do not remove them. Correlated inputs also need a suitable model. A more precise instrument is useful only if the actual measurement evidence supports the smaller uncertainty.
Try the decision yourself: which assumption would you check before recommending the second method? Explain how it could change the concentration comparison. This exercise supports your own evaluation; it does not supply a laboratory procedure or a complete investigation.
Bring an unknown into the calibration range
In this simulated example, a student wants to estimate the concentration of one dye in water. At a fixed wavelength and path length, the Beer–Lambert relationship motivates testing a linear response. All standards use the same solvent and measurement settings as the sample. The numbers below are made up for practice.
| Dye (mg/L) | Absorbance | Residual |
|---|---|---|
| 0 | 0.005 | +0.0006 |
| 2 | 0.119 | −0.0032 |
| 4 | 0.242 | +0.0020 |
| 6 | 0.361 | +0.0032 |
| 8 | 0.473 | −0.0026 |
| 10 | 0.542 | −0.0514 |
| 12 | 0.591 | −0.1202 |
Plot all seven standards first. The response flattens at the higher concentrations. Fit absorbance A against concentration c for the five lower standards from 0 to 8 mg/L, keeping the higher points visible for comparison. The least-squares line for those points is A = 0.0589c + 0.0044, with c in mg/L. A residual is the measured absorbance minus the value predicted by the fit. The residuals from 0 to 8 mg/L are small on both sides of zero. At 10 and 12 mg/L they are much larger and both negative. Repeat those high standards in a real pilot before deciding where to limit the fit. The pattern shows a departure from the line in this dataset. It does not identify the cause.
The undiluted sample gives A = 0.567, above the highest absorbance in the fitted range. In the simulation, the student dilutes a 10.00 mL aliquot to a final volume of 20.00 mL using the same solvent and reads an absorbance of 0.328. This lies between the 4 and 6 mg/L standards.
cdiluted = (0.328 − 0.0044) ÷ 0.0589 ≈ 5.49 mg/L
coriginal = cdiluted × (20.00 ÷ 10.00) ≈ 11.0 mg/L
Three readings from that same diluted preparation, 0.324, 0.328 and 0.332, give original concentration estimates from about 10.85 to 11.12 mg/L. This spread describes repeat readings of one preparation. To assess the uncertainty of the concentration estimate, also consider the calibration fit, standard preparation and dilution volumes. Repeat the dilution separately if you need to test how much preparation contributes. The illustrative fit also does not establish a lower quantitation limit.
Carry precision through the calculation
Retain enough precision in intermediate working to avoid rounding away useful information. Present the final value and uncertainty consistently, with units and suitable decimal places under your class convention. Explain calculations so another reader can see how raw values produced the reported quantity.
- Record instrument information when measurements are collected.
- Apply the chosen uncertainty method consistently.
- Include uncertainty in relevant processed results and graphs.
- Avoid interpreting small differences as decisive when the evidence does not support them.
Use uncertainty to qualify the conclusion
Compare your finding with an appropriate reference or model while considering the uncertainty and method limitations. A discrepancy may require discussion of sample composition, assumptions or a possible bias, rather than simply more repeated trials. Explain which improvement would reduce the uncertainty or strengthen the interpretation, how it would work and whether it is practical with school resources. Keep that reasoning connected to your research question.
Still choosing a measurable response? Compare the Chemistry investigation ideas and pilot checks before planning your own measurements.
Your next-step checklist
Checklist choices stay in this browser.
Official references and measurement notes
Both simulated examples use original IBvia teaching data. The IB guide sets the assessment context. NIST and Eurachem explain measurement uncertainty and calibration practice.

