Between-Class Variability
Definition
The spread of measurements between items from different sources or categories. A forensic method is most discriminating when between-class variability is large relative to within-class variability, so that same-source items cluster together and different-source items separate clearly.
- Concept
- Variation between different source groups
- Contrast
- Within-class (same-source) variability
- Desired pattern
- Large relative to within-class variability
- Field
- Statistical evaluation of forensic evidence
Common questions
Why does a method need large between-class variability to be useful?+
If items from different sources vary about as much as items from the same source, a measurement cannot reliably tell same-source and different-source items apart, undermining any comparison conclusion.
Is between-class variability the same as discriminating power?+
They are related but not identical. Discriminating power also depends on how tightly items from a single source cluster together, so both between-class and within-class variability need to be weighed together.
Related terms
- Accuracy
- The degree to which a measured value agrees with the accepted true or reference value of the quantity. A biased instrument can...
- Measurement Error
- The discrepancy between a recorded value and the true value of the property being measured. Arises from instrument calibration limits, analyst technique,...
- Precision
- The degree to which repeated measurements of the same specimen under the same conditions agree with each other. Quantified by the standard...
- Repeatability
- The agreement between successive measurements of the same specimen made by the same analyst, on the same instrument, in the same laboratory,...
- Within-Class Variability
- The natural spread of measurements among items that share a common source, class, or category. For example, the range of refractive index...