Search

The generalized lambda distribution family fits the probability distributions of a wide variety of data sets, including the most important distributions encountered in the measurement applications (normal, uniform, Student's t, U-shaped, exponential). This paper illustrates how the four parameters needed for such distribution can be exploited in the expression of measurement uncertainty and to extend the information related to a measurement. The obtained representation allows an immediate calculation of coverage intervals and is particularly useful to support the techniques commonly applied in the estimation of the combined uncertainty. Moreover, in order to include the classical measurement information, a novel parameterization of the distribution is proposed. [ABSTRACT FROM AUTHOR]