The Structure of Objectivist Studies - Evaluation Methods in Biomedical Informatics - page 89

TABLE 4.1. Attribute-object class pairs.
Attribute Object class

Speed Information resource

Blood pressure Patient

Correct diagnosis Patient

New admissions per day Hospital ward team

Computer literacy Person

Number of base-pairs DNA strand

Instruments
The instrument is the technology used for measurement. The “instrument”
encodes and embodies the procedures used to determine the presence,
absence, or extent of an attribute in an object. For studies in biomedical
informatics, instruments include self-administered questionnaires, tradi-
tional biomedical devices (e.g., an image acquisition device), tests of
medical knowledge or skills, performance checklists, and the computer
itself, through logging software that records aspects of resource use. It is
apparent from these examples that many measurements in informatics
require substantial human interpretation before the value of an attribute is
inferred. A radiograph must be interpreted; performance checklists must be
completed by observers. In such instances, a human “judge,” or perhaps a
panel of judges, may be viewed as an essential part of the instrumentation
for a measurement process.

Observations
An observation is a question or other device that elicits one independent
element of measurement data.* As measurement is customarily carried out,
multiple independent observations are employed to estimate the value of
an attribute for an object. This is because multiple independent observa-
tions produce a better estimate of the “true” value of the attribute than a
single observation. Use of multiple observations also allows for the deter-
mination of how much variability exists across observations, which is
necessary to estimate the error inherent in the measurement. As will be
detailed in the next chapter, multiple observations sometimes are obtained
by repeated measurements conducted under conditions that are as close to
identical as possible. Under other circumstances, multiple observations are
obtained under very carefully varied conditions. For example, the “speed”
of an information resource can be assessed by calculating the average
time taken to perform a range of appropriately selected computational

* When observations are recorded on a form, the term “item” is often used to
describe a question or other probe that elicits one independent element of infor-
mation on the form.