AGGREGATION 2

Charles François (2004). AGGREGATION 2, International Encyclopedia of Systems and Cybernetics, 2(1): 65.

Collection	International Encyclopedia of Systems and Cybernetics
Year	2004
Vol. (num.)	2(1)
ID	◀ 65 ▶
Object type	Methodology or model

The compression of a set of data into less numerous units which presents their common properties.

The most radical aggregation mode is statistics, a powerful variety reductor.

H. SIMON notes that, in many complex systems, analysis is much simplified by the fact that the influences on each element of all other elements are organized in a simple way. This is because the anonymous character of the specific interactions leads to a very important simplification, not only of the global system, but also of the functions which describe the component systems (1990, p.138).

While such a simplification is very useful, it also may be altogether sometimes questionable, because it may imply (and generally does) a loss of information. Furthermore, it is the observer who decides the level of aggregation, in accordance with his/her specific interest (and personal blindspots). This level can be for instance micro- or macrophysical, chemical, biological or sociological.

Another difficulty. has been signaled by K. KRIPPENDORFF: “Aggregation leads to misleading indicators and theories whenever the whole collection exhibits an organization not expressed in a mere summation” (1986, p.1)

In complex systems, aggregation of data may occult relevant interrelations… or even introduce imaginary ones, when the “common” properties result merely from a mere unconscious and unwarranted hypothesis. These distortions may occur, for example, in the construction of complex computer models, needed in systems dynamics.

This problem turns worse if the aggregated variables do not any more correspond to observable state variables, as for instance in the global aggregate called “pollution”. J. MARTINO observes: “… the effect of pollution on population actually occurs as effects of specific pollutants on specific individuals” (1979, p.17).

See also