Measurement, Uncertainty, and Error Modeling
Measurement as an Inference Problem
Why Measurement Is Never Perfect
Every measurement is an estimate of an underlying quantity through an instrument, protocol, and model. Advanced science treats measurement as an inferential chain: the target quantity, the sensor response, the calibration model, and the uncertainty budget all contribute to the final value.
Practical Insight
A smaller standard deviation does not necessarily imply better measurement if the estimate is biased.
Which quantity best characterizes the spread of repeated unbiased measurements?
Standard deviation describes dispersion around the mean for repeated measurements.
Correct answer: Standard deviation
Name the type of error that shifts all measurements in the same direction.
Systematic error produces consistent deviation from the true value.
Correct answer: Systematic error
Error Vocabulary
Common Uncertainty Sources
| Source | Typical Consequence |
|---|---|
| Instrument drift | Slow change in readings over time |
| Sampling noise | Random scatter around the mean |
| Operator variation | Differences across observers or labs |
| Model misspecification | Underestimated uncertainty or biased inference |
If repeated measurements are tightly clustered but far from the true value, the data are:
Tight clustering indicates precision; distance from truth indicates inaccuracy.
Correct answer: Precise but inaccurate
What is one purpose of calibration?
Calibration reduces systematic error by mapping readings to standards.
Correct answer: To relate instrument output to known reference values.
Building an Uncertainty Budget
-
1
List all known sources of uncertainty.
-
2
Classify them as random, systematic, or model-based.
-
3
Assign magnitudes using data, standards, or expert judgment.
-
4
Propagate uncertainties through the measurement model.
-
5
Report final values with transparent assumptions and limits.