8127/frame/fm

 
i.
Signal vs. noise.
Closely related to the scale of detection is the issue of
signal vs. noise. This type of problem manifests itself more in analyses involv-
ing chemical interactions or instrumental data output than in strict visual
comparison. Although noise may certainly be present in visual images, it is
more likely to result simply in a loss of data or resolution rather than be
misinterpreted as a true trait. The distinction of signal from noise, when an
instrument or process stands between the evidence and our eyes, requires an
additional level of consideration.
Most people are familiar with the blue dots that represent the final result
of one kind of DNA typing using PCR. These blue dots are the culmination
of a chain reaction of chemical interactions instigated by the hybridization
of specific DNA sequences. It is a well-known fact among practitioners that,
shortly after representative results are achieved, nonspecific background color
begins to appear. It would be very easy for someone unfamiliar with the test
system to be misled and become concerned about whether the faint signals
represent DNA types actually present in the sample. For this reason it is
standard practice to document the dot pattern with a high-quality photograph
* Although such an instrument does exist, we acknowledge that this would be an unusual
endeavor.
8127/frame/ch06 Page 128 Friday, July 21, 2000 11:47 AM

---

Classification, Identification, and Individualization
129
immediately after optimal development and to discard the original strips
which, over time, no longer represent the true genetic result.
Similar decisions must be made for any kind of chromatographic or
electrophoretic analysis. Are the blips along the baseline random fluctuations
inherent in the system or do they represent a minor component that could
be of interest? The ability to distinguish signal from noise depends on under-
standing both the nature of the evidence and the nature of the test system.
To make these crucial decisions reliably, the analyst relies on extensive vali-
dation of both the test system and instrumentation, as well as her own
education, training, and experience. This is particularly critical for samples
of a forensic nature.