8127/frame/fm

a.
Frequency Estimates
The simplest and most common method of expressing the strength of the
evidence is through frequency estimates. Information is gathered about how
common any item or trait might be in the population of possible sources.
This frequency is then used to convey how unusual it might be to encounter
this evidence by chance alone. In biological evidence, this has been called the
random match probability. It simply estimates the probability of seeing this
evidence if it were from a source other than the reference in the instant case.
For biological evidence this is rather simple, while for other evidence types,
such as trace, it has been called impossible.
For any biological evidence, surveys of various populations are per-
formed, and the results tabulated and tested according to standard genetic
theory. This procedure provides reasonable estimates for the probability of
encountering a biological evidence profile at random (Griffiths et al., 1993).
Population studies performed on human populations rely on the relative
stability of human populations, our knowledge about their distribution, and
our ability to acquire a representative sampling. Using the laws of genetics,
we can also test for the independence of traits.
However, for inanimate evidence, in particular trace evidence, frequency
studies provide much more ephemeral data. Populations of manufactured
materials change both quickly and unpredictably; a frequency survey repre-
sents a snapshot in time. The composition of these populations also depends
greatly on geographical location and may be extremely localized. Therefore,
it is difficult to know if the sample fairly represents the population (Horrocks
et al., 1999) or if the population is the correct one to estimate the strength
of some evidence found at another time or location. Because of this uncer-
tainty, errors associated with frequency estimates could be higher than the
estimates themselves. It is difficult to answer such a question as “What is the
probability of seeing this evidence by chance alone?” when the frequency is
a moving target. Finally, and as a direct consequence of the preceding issues,
it is difficult to imagine how to test for independence between two inanimate
objects because no physical laws govern their distribution or dispersal. When
is it appropriate to multiply traits in estimating the frequency of a particular
type of evidence (Gaudette, 1978; 1982; Bodziak, 1990)? Can we derive a
combined estimate of the strength of the class of evidence using different
types of the same class of evidence (Deadman, 1984a,b; Deadman 
in
Safer-
stein, 1998), or even two different traits from the same evidence such as
mitochondrial DNA and microscopic traits (Shields, 1998)?