Sources of Uncertainty (from

Morgan and Henrion)

Jake Blanchard
Spring 2010

Uncertainty Analysis for Engineers 1

Types of events
 Subjective probability distributions are only
suitable for certain types of events
 Empirical Quantities – measurable
properties of real-world systems
 Constants – fundamental physical constants
(certain by definition)
 Decision Variables – quantities over which
the decision maker exercises direct control
 Value Parameters – aspects of the
preferences of decision makers (eg. risk
tolerance or value of life)

Uncertainty Analysis for Engineers 2

Event Types (cont.)
 Index Variables – identify a location or cell in
the spatial or temporal domain (eg. a
particular year or geographical grid)
 Model Domain Parameters – specify domain
or scope of system (eg. Last year modeled,
spatial extent of model, etc.)
 State Variables – minimal subset of all
variables from which all other variables can
be calculated
 Outcome Criteria – variables used to rank
outcomes

Uncertainty Analysis for Engineers 3

 Sources of Uncertainty – Empirical
Quantities
 Statistical variation – random error in direct
measurement
 Systematic error – difference between true
value of a measured quantity and mean of
measured values
 Linguistic imprecision – (“Pat is tall” vs. “Pat is
over 6 feet tall”)
 Variability – (eg retail price of gasoline or flow
rate of a river)
 Randomness – variation that cannot be
attributed to a pattern or model (function of
available knowledge)
Uncertainty Analysis for Engineers 4
 Uncertainty About Model Form

 If we pick wrong model (eg normal vs. beta

distribution), errors will result

Uncertainty Analysis for Engineers 5