8127/frame/fm

 
B.
Validation
In contrast to optimization, which concentrates on the test system, 
 
validation
concentrates on the samples themselves. In performing validation studies,
we ask what quantitative and qualitative properties of the samples themselves
might confound our ability to get reliable results or to get results at all. We
intentionally perturb the samples to determine at what point no result, an
ambiguous result, or even an incorrect result might be produced. Because of
this, validation is particularly critical for systems intended for forensic use
where the samples tested will not be of uniform quality and quantity.
8127/frame/ch09 Page 223 Friday, July 21, 2000 11:44 AM

---

224
Principles and Practice of Criminalistics
The purpose of validation is to establish limits within which we are confi-
dent of the results obtained. Validation also helps us to establish the variance
inherent in the system as a whole, and to gain a further understanding of one
kind of explainable difference. For example, if a sample is tested 100 times, the
results will look similar, but small differences will undoubtedly manifest. We
learn how much variation may exist between tests, so that we may determine
criteria within which we can either conclude that two different samples may
share a common origin or exclude them as having a common origin. How
sensitive is the system to the quantity of sample? What are the lowest and highest
quantities that can reliably be analyzed? What artifacts begin to appear at the
extreme ends of the spectrum and can we use them as diagnostics? What do
mixtures of similar substances look like in a particular test system, and are there
indicators that might alert us to the presence of a mixture? How might various
environmental factors to which forensic samples are commonly subjected influ-
ence the ability to analyze the samples and to interpret the results obtained?
These are the kinds of questions that validation studies attempt to answer.