Introduction
Chemometrics applies statistics and data analysis to chemical measurements, ensuring that experimental results are reported and interpreted honestly — a skill relevant to every laboratory-based chapter in this subject, especially analytical separations (Chapter 7) and spectroscopy.
1. Accuracy, precision, and error
Accuracy is how close a measurement is to the true value; precision is how close repeated measurements are to each other, regardless of whether they are accurate. A measurement can be precise but inaccurate (consistently off-target due to a systematic error) or accurate on average but imprecise (scattered around the true value due to random error). Systematic errors (from instrument calibration, technique) shift all measurements in the same direction and can, in principle, be corrected; random errors (from unavoidable small fluctuations) scatter measurements in both directions and are reduced (though never eliminated) by averaging more measurements.
2. Significant figures and uncertainty
Significant figures communicate the precision of a measurement; reporting more digits than the measuring instrument justifies falsely implies greater precision than actually exists. Every measured quantity should be reported with an appropriate number of significant figures and, where relevant, an explicit uncertainty (e.g., 12.4 ± 0.1 mL) — a habit examiners and real laboratories both expect.
3. Basic descriptive statistics for chemical data
The mean summarises a data set’s central value; the standard deviation quantifies the spread (precision) of repeated measurements around that mean. These are the minimum tools needed to describe a set of replicate measurements meaningfully rather than reporting a single number as if it were exact.
4. Calibration and signal–concentration relationships
Most analytical instruments do not measure concentration directly but a related signal (absorbance, current, peak area). A calibration curve is built by measuring the signal for several samples of known concentration, typically producing a linear relationship (signal = m × concentration + c) that can then be used to determine the concentration of an unknown sample from its measured signal.
5. Quality of analytical results
Reporting analytical results honestly means combining all of the above: acknowledging measurement uncertainty, using an appropriate and validated calibration, and distinguishing genuine chemical differences between samples from mere measurement noise — a discipline as important to real research and industrial quality control as it is to answering examination questions correctly.
Key Takeaways
- Accuracy (closeness to true value) and precision (closeness of repeated measurements to each other) are independent qualities; systematic and random errors affect them differently.
- Significant figures and stated uncertainty communicate a measurement’s actual precision honestly.
- Mean and standard deviation are the essential descriptive statistics for any set of replicate chemical measurements.
- A calibration curve converts an instrument’s raw signal into a usable concentration value, and is only reliable within the concentration range over which it was actually constructed.