a measurement from a meaningless assignment of numbers. We offer an alternative
definition that applies to both the process and the result:
Measurement (alternative definition):
The symbolic representation of a state, event or attribute to aid in the making of a
decision.
This definition highlights three important aspects of measurement not apparent from
the first two definitions:
(1) The results of measurement need not be numeric: grade A, red and sodium are all
legitimate measurement results in the appropriate context. One of the most valuable
aspects of symbolic representation is that we use the symbols in our models
to make predictions. Mathematical models and numeric symbols particularly help
to quantify predictions that might otherwise be qualitative or subjective.
(2) Every measurement has a purpose. This is the distinction we seek between a
meaningful measurement and meaningless assignment of numbers. In a great
many measurements, especially outside the calibration laboratory, the purpose
influences the design and outcome of the measurement. Consequently, measurement
results may have meaning only within the context of that purpose. Results
used for other purposes or gathered without purpose are potentially dangerous.
(3) Decisions are associated with risks and rewards. This highlights the need to
know the uncertainty in a measurement in order to assess the risk or optimise the
measurement.
These three points are the main themes of Sections 1.2.2 to 1.2.4.
Exercise 1.1
Consider some of the measurements you make.What decisions do they influence?
What are some of the risks and rewards associated with those decisions?
1.2.2 Measurement scales
In this section, we look at some of the basic systems for the symbolic representation
of measurement results: our measurement scales. Table 1.1 provides a coarse summary
of the various types of measurement scale that we use. The classification given in the
table is based on the algebraic properties underlying the measurement scales. While
there are other systems of classification we have chosen this one in order to highlight
the fundamental limitations associated with the different systems of representation.
Nominal scales
Measurements on nominal scales are essentially a naming, and often the results do not
look like measurements at all. An example familiar to most is the number on a football
player’s shirt. In many sports, the number on a player’s shirt indicates the position the