In the interpretation of results from a particular technique, one often must evaluate the performance of the measurement on the basis of objective measures that are called figures of merit. The following table provides a description of the more common figures of merit used in analytical chemistry. Even more important than the evaluation of performance is the validation of metrology and the assessment of validation. The same terms used as figures of merit are critical in the validation process. Because many standardization bodies issue their own definitions regarding these terms, it is not practical to be inclusive; the reader must be aware that other terms may sometimes be substituted for those used here.
Parameter or metric | Definition | Comments |
Accuracy | The deviation of a measured value from the true value (τ), which cannot typically be determined, but which is often approached by use of a standard method under standard conditions. | Can be expressed as: accuracy = (xav – τ)/τ, where xav is the mean of a series of measurements. |
Precision | The reproducibility of a series of replicate measurements, typically under the same operating conditions in a relatively short period of time. | High precision does not imply high accuracy; reproducibility and repeatability often are used interchangeably, but the former typically refers to precision between two instruments or laboratories. |
Sensitivity | Typically the ability to distinguish between small differences in concentration between samples at a desired confidence level. | Often specified by response to a de facto or consensus standard; a simple measure is the slope of a calibration curve adjusted for recovery in the sampling process. |
Limit of Detection (LOD) | The lowest measureable concentration of an analyte by use of a particular metrology at a desired confidence level. | Typically defined as a response that is some multiplicative factor of the noise level; specified signal-to-noise ratios of 2 to 4 are typical; as with sensitivity, must adjust for recovery in sample preparation processes. |
Linear Dynamic Range (LDR) | The linear range of a calibration curve obtainable with a particular metrology and analyte. | Within the LDR, the difference in response for two concentrations of a given compound is proportional to the difference in concentration of the two samples. |
Selectivity (α) | The ability to distinguish the signal of an analyte from the signal of interferences, the matrix or degradants. | Typically expressed as a ratio, such as the ratio of capacity factors in chromatography; sometimes called specificity. |
Analysis Speed | The time required for sample preparation and measurement. | Sometimes necessary to include data processing time. |
Throughput | The number of samples that can be run in a given time. | Related to ease of automation, below. |
Ease of Automation | The ability of a measurement to be performed without operator attention. | Related to throughput, above, because high throughput typically requires automation and minimal operator attention. |
Ruggedness | The response of an instrument or technique to adverse conditions. | Adverse conditions include extremes in temperature, humidity, dust; includes rough handling; closely related to reproducibility, because it carries similar information content; distinguished from robustness, which is the ability of an instrument to remain unaffected by a small but intentional change in operating parameter, and still perform within specification. |
Portability | The ability to use an instrument in other than a fixed location or installation. | Can include the ability to use a device in the field in addition to the laboratory. |
Environmental Acceptability | The efficiency in terms of low waste generation or low power consumption. | Also described as sustainability, and often focused on the minimization of use of hazardous substances. |
Economics | The sum of costs required to operate the sample preparation, analysis and data processing steps of a measurement. | Typically includes the costs of equipment, supplies, labor, utilities, and insurance; labor cost must consider the skill level required of personnel. |