Biomedical Engineering Reference
In-Depth Information
situations, and scenarios, so that what occurred will never
be completely repeated again. Every cubic centimeter of
soil is different from every other cubic centimeter. The
same goes for a sample of water, sediment, air, and organic
tissue. And, these all change with time. Taking a sample in
the winter is different from that in the summer. Condi-
tions in 1977 are different in so many ways from condi-
tions in 2007. And, of course, there are errors. Some are
random in that the conditions that led to the cases in this
topic are partially explained by chance and things that are
neither predictable nor correctable, although we can ex-
plain (or at least try to explain) them statistically, (e.g.,
with normal distributions).
Other error is systematic, such as those of my own
bias. I see things through a prism different from anyone
else's. This prism, like yours, is the result of my own
experience and expertise. This prism is my perception
of what is real and what is important. My bias is heavily
weighted in sound science, or at least what I believe to
be sound science (as opposed to ''junk science'').
41
Sound science requires sufficient precision and accuracy
in presenting the facts. ''Precision'' describes how re-
fined and repeatable an operation can be performed,
such as the exactness in the instruments and the
methods used to obtain a result. It is an indication of the
uniformity or reproducibility of a result. This can be
likened to shooting arrows,
42
with each arrow repre-
senting a data point. Targets A and B in
Figure 8.2-7
show equal precision. Assuming that the center of the
target; that is, the bull's-eye, is the ''true value,'' data
set B is more accurate than A. If we are consistently
missing the bull's-eye in the same direction at the same
distance, this is an example of bias or systematic error.
The good news is that if we are aware that we are
missing the bull's-eye (e.g., by comparing our results
with those of known standards when using our analytical
equipment), we can calibrate and adjust the equipment.
To stay with our archery analogy, the archer would move
her sight up and to the right.
Thus, ''accuracy'' is an expression of how well a study
conforms to some defined standard (the true value). So,
accuracy expresses the quality of what we find, while
precision expresses the quality of the operation by which
we obtained our finding. So, the other two scenarios of
data quality are shown in targets C and D. Thus, the four
possibilities are that our data is precise but inaccurate
(target A), precise and accurate (target B), imprecise
and inaccurate (target C), and imprecise and accurate
(target D).
At first blush, target D may seem unlikely, but it is
really not all that uncommon. The difference between
targets B and D is simply that D has more ''spread'' in the
data. For example, the variance and standard deviation
of D is much larger than that of B. However, their
measures of central tendency, i.e., the means, are nearly
the same. So, both data sets are giving us the right
answer, but almost all of the data points in B are near the
true value. None of the data points in D is near the true
value, but the mean (average location) is near the center
of the bull's-eye; so, it has the same accuracy as target B,
but with much less precision. The key is that precision
and accuracy of the facts surrounding a case must be
known.
Even if we agree on the facts, we all will not agree on
which of the virtues and vices are best or even whether
something is a virtue or a vice (e.g., loyalty). But one
concept does seem to come to the fore in most major
religions and moral philosophies: empathy. Putting
oneself in another's situation is a good metric for virtu-
ous acts. The ''Golden Rule'' is at the heart of Immanuel
Kant's ''categorical imperative.'' With apologies to Kant,
here is a simplified way to describe the categorical im-
perative: When deciding whether to act in a certain way,
ask if your action (or inaction) will make for a better
world if all others in your situation acted in the same
way. An individual action's virtue or vice is seen in
a comprehensive manner. It is not whether one should
cut corners and use deficient materials when designing
a device, it is whether everyone in the same situation
should do likewise. A corollary to this concept is what
Elizabeth Kiss, formerly of Duke's Kenan Center for
Ethics and now president of Agnes Scott College calls
the ''Six O'clock News'' imperative. That is, when de-
ciding whether an action is ethical or not, consider how
my friends and family would feel if they heard about all
of its details on tonight's TV news. That may cause me to
consider more fully the possible externalities and con-
sequences of my decision (and maybe even tempt me to
''overdesign'').
A
B
C
D
Psychological aspects of ethics
Figure 8.2-7 Precision and accuracy. The bull's-eye represents
the true value. Targets A and B demonstrate data sets that are
precise, targets B and D demonstrate data sets that are accurate,
and targets C and D demonstrate data sets that are imprecise.
Target B is the ideal data set, which is precise and accurate.
Engineers are familiar with conditions and constraints.
For example, we know from chaos theory the importance
of initial conditions on outcomes, but also know these can
