The Structure of Objectivist Studies - Evaluation Methods in Biomedical Informatics - page 93

healthcare providers, the investigator may want to know each respondent's
years of professional experience: naturally, a ratio measure. Frequently,
however, such attributes are assessed using discrete response categories,
each containing a range of years. Although this measurement strategy
provides some convenience and possibly some sense of anonymity for the
respondent (and may, in some cases, generate more complete data with
fewer missing values), it reduces to ordinal status what is naturally a ratio
variable, with inevitable loss of information. Even if the data are later going
to be categorized when eventually analyzed and reported, collecting and
storing the data at the highest level of measurement is a safer strategy. Data
can always be converted from higher to lower levels of measurement, but
it is not possible to go the other way.
Self-Test 4.2
Determine the level of measurement of each of the following:
1. A person's serum potassium level;
2. A health sciences center's national ranking in grant funding from the
National Institutes of Health;
3. The distance between the position of an atom in a protein, as predicted
by a computer model, and its actual position in the protein;
4. The “stage” of a patient's neoplastic illness;
5. The internal medicine service to which each of a set of patients is
assigned;
6. A person's marital status;
7. A person's score on an intelligence test, such as an IQ test.
Importance of Measurement
Having introduced some concepts and terms, the importance of measure-
ment in objectivist studies can be appreciated by revisiting the major
premises underlying the objectivist approaches to evaluation. These
premises were originally introduced in Chapter 2, but they are re-stated
here in a somewhat revised form to exploit this new terminology. Like all
premises, these are based on assumptions that reflect idealized views of the
world and our ability to understand it through certain methods of empiri-
cal research. Readers who have difficulty accepting these assumptions
might find themselves more attracted to the subjectivist approaches dis-
cussed in Chapters 9 and 10.
In objectivist studies, the following can be assumed:
• Attributes inhere in the object under study. Merit and worth are part
of the object and can be measured unambiguously. An investigator