All uncertainty contributors must have their impact quantified and be expressed as standard uncertainties, otherwise combination of the uncertainty contributors cannot be achieved in the budget. Assessments use Type A and Type B evaluations depending on the contributing factor and data available. Results are expressed as either:
- Experimental Standard Deviation (Type A)
- Equivalent Standard Deviation (Type B)
These provide a value for the standard uncertainty as determined by the standard deviation.
It is important to consider what probability distribution best describes the shape of the standard deviation or data available as this will impact on any necessary transformation of the data to make the standard uncertainty comparable between contributors.
Data described by the following distributions are transformed by dividing by their respective denominators:
Normal distribution: divided by 1 – effectively the standard deviation is equivalent to the standard uncertainty
Rectangular/Uniform distribution: The range of the uncertainty is divided by the square root of 3. As such the result is equivalent to the standard uncertainty.
Triangular distribution: The range of the uncertainty is divided by the square root of 6. As such the result is equivalent to the standard uncertainty.
These are the most commonly encountered distributions in pathology.