100
9
Probability and Likelihood
proposition that a necessary condition for an event to be causally conditioned is that
it can be predicted with certainty. If, however, we compare a prediction of a physical
phenomenon with more and more accurate measurements of that phenomenon, one is
forced to reach a remarkable conclusion—that in not a single instance is it possible
to predict a physical event exactly, unlike a purely mathematical calculation. The
“indeterminists” interpret this state of affairs by abandoning strict causality and
asserting that every physical law is of a statistical nature. The opposing school of
“determinists” asserts that the laws of nature apply to an idealized world-picture,
in which phenomena are represented by precise mathematical symbols, which can
be operated on according to strict and generally agreed rules and to which precise
numbers can be assigned (to which an actual measurement can only approximate);
in the corresponding mentally constructed world-picture, all events follow definable
laws and are strictly determined causally; the uncertainty in the prediction of an event
in the world of sense is due to the uncertainty in the translation of the event from the
world of sense to the world-picture and vice versa. It is left to the interested reader
to pursue the implications with respect to quantum mechanics (with which we shall
not be explicitly concerned in this book).
Sommerhoff (1950) formulated probability in the following terms: Given a system
whose initial state can be one of a setupper QQ ofnn alternativesupper Q 1 comma upper Q 2 comma ellipsis comma upper Q Subscript n BaselineQ1, Q2, . . . , Qn, of which
a certain fraction m divided by nm/n will lead to the subsequent occurrence of an event upper EE that
is to be expected in the normal development of the system, then the probability
that any particular member of upper QQ leads to upper EE is given by the fraction m divided by nm/n. Note
that this formulation only applies to the effects of the initial states, not to the states
themselves. It has the advantage of avoiding any assumption of equally probable, or
equally uncertain, events.
Before any further discussion about probability can take place, it is essential to
agree on what is meant by the possible results from an experiment (or observa-
tion). These results are called “events”. Very often abstract models, corresponding
to idealized events, are constructed to assist in the analysis of a phenomenon.
9.2
Fundamentals
The elementary unit in probability theory is the event. One has a fair freedom to define
the event; simple events are irreducible and compound events are combinations of
simple events. For example, the throw of a die to produce a 5 (with probability 1/6)
is a simple event, and combinations of events to yield the same final result, such as
three 2s, or a 5 and a 1, are compound events. Implicitly, the level of description
is fixed when speaking of events in this way; clearly, the “event” of throwing a 6
requires many “sub-events” (which are events in their own right) involving muscular
movements and nervous impulses, but these take place on a different level.