Part
One: Probability
Probability
Perspectives
Poincare
and Chance
"A very small
cause which escapes our notice determines a considerable effect that we cannot
fail to see, and then we say that that effect is due to chance."1
Chance can operate at
several levels, based upon the nature of that which escapes our notice. In a
pragmatic sense, chance is usually viewed at two levels: unobserved factors
which could be observed or measured; and irreducible uncertainty. We can have a
very naïve sense of chance, in which we claim no prior knowledge. We can modify
this naïve sense of chance if we can claim some basic or prior knowledge. The
basic idea of chance is that it is driven by the unknown (and unknowable)
forces behind a process. Probability views chance from a hypothetical
standpoint, while statistics views probability from an empirical/experiential
standpoint.
1.
H.
Poincare, Science and Method, translated by F.Maitland, pp. 67-68.(New
York: Dover Publications, 1952).
Prediction and a Fair Die
Randomness embodies the
uncertainty in the prediction of outcomes in selected processes. Consider the
fair, six-sided die (d6). A fair d6 has six, equally likely faces, usually with
face values {1,2,3,4,5,6} or more commonly with spot-groups in place of
numbers. Consider a thought experiment in which we predict the face that will
show prior to tossing the d6. Then check the prediction against the actual
toss. Try this a few times. Why can't we predict the tosses reliably?
Probability
forms the theoretical foundation for Statistics.
Probability
underlies Statistics like Algebra underlies Calculus. While some concepts in Statistics
do not relate directly to Probability, Statistics is in many ways an extension
of Probability. So we study Probability first.
Some
loose definitions:
…from Webster's New Collegiate
Dictionary… Probable: supported by evidence strong enough to
establish presumption but not proof. Probability: the relative degree to
which an event is probable.
…from Statistics
(Freedman, Pisani, Purves and Adhikari)…Chance: the percentage of
time something is expected to happen, when the basic process is repeated
independently and under the same conditions.
The basic principle of
Probability is that of a likelihood - in fact, the term probability is
often used as a synonym or alias for likelihood. Imagine a black box with a red
button in front. Each time the button is pressed, something simple happens. The
box is in the same state each time - the box resets after each cycle.
A Black Box
Imagine
that a team is charge of learning about the box. This team is not given any
detailed information about the box, and
they are not allowed to harm or damage the box - all they can do is
press the button and note the results.
One
person pushes the button - over and over and over and over again, ad infinitum.
Each time the button is pressed, another person writes down what happens, that
is, what the box does when the button is pressed. A third person summarizes
what the second person wrote.
After
a large number of box-cycles (press button and see what happens), the team can
gain some insight into the box 's workings.
Specifically, the team
can say something about the following:
What sort of things can the box do when the button is pushed ?
How often does the box seem to do certain specific things ?
The second item is at
the heart of Probability - in long runs of activity, how often can we expect
specific results from the box? This is the idea of likelihood - on average, how
often do we expect a particular event to occur ?