8127/frame/fm

 
scientific method
provides a framework for hypothesis testing.
In reality, we can never prove that an idea, concept, or theory is true — we
can only fail to prove that it is false. In the absence of information that a
theory is untrue or incorrect, we accept it as correct until new information
is obtained that demonstrates otherwise. Both in science and forensic science,
we frequently have an idea in mind — this bullet came from that gun. This
8127/frame/ch01 Page 5 Friday, July 21, 2000 11:32 AM

---

6
Principles and Practice of Criminalistics
is called the 
 
null hypothesis
* (Fisher, 1949). While we cannot scientifically
prove our hypothesis, we can try very hard to disprove it. If we perform
discriminating and adequate testing, and repeatedly fail to disprove the null
hypothesis, we may become convinced that our original hypothesis is true —
that the bullet did pass through the gun. If the testing does, in fact, disprove
the null hypothesis, we must accept the alternate hypothesis (Neyman and
Pearson, 1928) — that the bullet was not fired from the gun. Although this
idea is a basic tool of the working criminalist, it is rarely recognized or
articulated. Another way of examining hypotheses is in a Bayesian framework
where competing hypotheses are compared and their relative likelihood cal-
culated (Evett, 1983; Taroni et al., 1998). As we will see, both logical frame-
works find a use in forensic science, and one may be more useful than another
at various stages in the process.
An experimental result has no standing until it is disseminated to the
relevant scientific community for review. This is usually accomplished by
presentation at a scientific meeting or publication in a peer-reviewed journal.
Publication of the methods used in obtaining the data provides a chance for
other colleagues to repeat and reproduce the experimental result. Thus, sci-
ence is a product of the community, not individuals. It is much too easy to
be led astray without the foil of other scientists to refute one’s ideas. In basic
research, an experimental result is not accepted until it can be reproduced
by an independent worker (Maddock, 1989). This concept is followed in
forensic science when specific techniques are developed or adapted for foren-
sic use. Laboratories will, for instance, carry out collaborative studies to
confirm that a new method yields equivalent results in different hands.
In actual casework, however, duplication of testing is not always possible
for a variety of reasons, not the least of which may be an inherent limitation
in sample size. Thus, in forensic casework, the confirmation of work product
frequently takes the form of independent review, either by another analyst
in the laboratory or by an expert assisting opposing counsel. It is essential
that the work be reviewed both to catch and correct any clerical errors and
to establish that the conclusions are supported by the data. It is the obligation
of the forensic scientist to interpret the data objectively and to form and state
a conclusion regarding the results.