MEASUREMENT THEORY

IRA FIKA YUMI

Measurement is the acquisition of information about a state or phenomenon (object of measurement) in the world around us. This means that a measurement must bedescriptive (observable) with regard to that state or object we are measuring: there must be a relationship between the object of measurement and the measurement result.

Let us define pure science as science that has sole purpose of describing the world around us and therefore is responsible for our perception of the world. In pure science, we can form a better, more coherent, and objective picture of the world, based on the information measurement provides. In other words, the information allows us to create models of (parts of) the world and formulate laws and theorems. We must then determine (again) by measuring whether this models, hypotheses, theorems, and laws are a valid representation of the world. This is done by performing tests (measurements) to compare the theory with reality.

A scale is an organized set of measurements, all of which measure one property.

The types of scales reflect the types of measurements: 1. 2. 3. 4. 5. nominal scale, ordinal scale, interval scale, ratio scale, absolute scale.

Illustration: Levels of measurements (S. S. Stevens, 1946)

ABSOLUTE Abs. unit RATIO Abs. zero

INTERVAL Distance is meaningful

ORDINAL States can be ordered

NOMINAL States are only named

1. Nominal scale Examples: numbering of football players, detection and

alarm systems, etc.

Any one-to-one transformation can* be used to change the scale.
*Stevens did not say that transformations that are not 'permissible' are prohibited. http://mu.dmt.ibaraki.ac.jp/yanai/neu/faq/measurement.html#exmpls

1 0

1 0

1 0

1 0

1 0

1 0

OK

2. Ordinal scale Examples: IQ test, competition results, etc.

Any monotonically increasing transformation, either linear or nonlinear, can be used to change the scale.

OK

3. Interval scale Examples: time scales, temperature scales (C, F), etc., where the origin or zero is not absolute (floating).

Any increasing linear transformation can be used to change the scale.

+
D B C B A

OK
B A D C

D C

A B C D

A
2X+1

NB: x(-1) does not change the interval but does change the order: A>C.
8

4. Ratio scale Examples: temperature (K), distance, mass, current, voltage scales, etc., where the origin or zero is absolute.

The only transformation that can be used to change the scale is the multiplication by any positive real number.

OK
0 0 0

NB: x(-1) does not change the ratio and interval but does change the order.
9

x2

5. Absolute scale Examples:measurement of any physical quantities by comparison against an absolute unit (reference).

No transformation can be used to change the scale.

10

10

10

10

10

10

-10

-10

(Same interval)

(Same ratio)

(Same ratio, different order)

Not the same absolute values.

10

The reason is that the structure of the scale is left variant ,that is :

The rank order of the points is unchanged

The ratio of the points are unchanged The zero point is unchanged

11

Three measurement are discussed on the following pages

Fundamental measurements A fundamental is one where the numbers can be assigned to the property by reference to naturak laws and which does not depend on the measurement of any others variable. Derived measurements A derived measurement is one that depens on the measurement of two or more other quantitiees. It depends on the measurementsof both mass and volume. Fiat measurement Fiat measurements are those that relate numbers to properties of objects or events on the basis of arbitrary definitions.

Types of measurements
To represent a state, we would like our measurements to have some of the following characteristics. Distinctiveness: A B, A B. Ordering in magnitude: A < B, A B, A > B. Equal/unequal intervals: IA-BI < IC-DI, IA-BI IC-DI, IA-BI > IC-DI . Ratio: A k B (absolute zero is required). Absolute magnitude: A ka REF, B kb REF (absolute reference or unit is required). These five characteristics are used to determine the five types (levels) of measurements.
13[1] Reference:

Sources of Error 1. Measurement operations states imprecisely 2. Measurer 3. Instrument 4. Environment 5. Attribute unclear

Reliability refers to the proven consistency of the measure, that is, the same ( or similar) measure would be reproduced by different observers. Accurate measure refers to the representation of the fundamental value of the property being measure.

 The two fundamental measures in accounting are capital and profit and they are both derived measures.
Capital is derived from transactions and revaluations that occur in financial markets. Profit can be derived from mathcingof expenses with revenue or the change in the capital over period Capital can be defined and derived in various ways, including historical cost, operating, financial, or fair value, Profits should not be based on a notion of realisation.

Conclusion
The concept of scale type is an important one, and Stevenss terminology is often suitable. We must keep in mind, however, that scale types are not fundamental attributes of the data, but rather, derive from both how the data were measured and what we conclude from the data. To restrict our investigation only to hypotheses and calculations permitted by an a priori assignment of scale type would be far more irresponsible. Responsible data analysis must be open to anomaly if it is to support scientific advancement.

17