One Concept Equals One Data Element: A Rule for Developing Data Elements - High-Performance Web Databases: Design, Development, and Deployment - page 83

and valid values are restrictive, it is not as useful as one that serves many
needs in the organization. An example of an extremely useful data element
is a Social Security number. Any information system that stores personal
data can use this data element as the key to its personnel records with con-
fidence that innumerable other systems are using the same data element.

Making data elements self-evident is one of the data administrator's
main responsibilities. By constructing their data repository schema, data
administrators establish the attributes to be documented for each data ele-
ment. These attributes must be comprehensive enough to ensure that the
data elements can be understood completely and consistently. Sometimes,
data elements that are poorly named and not defined create no problems
for the long-time users of the system because they already have a complete
understanding. When systems are redeployed in a distributed environment
or are reengineered and enhanced, however, both the previous users and
the new users are likely to assign their own meaning to the new and revised
data elements that are poorly named and defined. This may lead to incor-
rect reports to management, poor decisions, misunderstandings, or
delayed business functions. Data administrators assist in the creation of
data elements by developing the set of attributes that provides the struc-
ture to facilitate comprehensive definitions.

Homogeneous valid values represent a set of like items.
The valid values are the representations, or instances, of the concept that
the data element embodies. Blue, green, and brown are all instances of eye
color. The way in which the valid values are like each other is the concept.
For example, it is helpful to ask what do blue, green, and brown have in
common? Because they all represent colors of people's eyes, the concept
is person eye color.

Homogeneity.

The presence of valid values that are not homogeneous indicates that
more than one concept is being included. The example in Exhibit 8 con-
tains valid values that are not homogeneous. These valid values represent
more than one concept. The following phrases can be analyzed to deter-
mine the concepts they represent:

• the amount of hair a person has
• the color of the hair that a person has
• whether or not a person has hair

The phrases do not take into account whether the person wears a wig or
whether the hair is more than one color (e.g., salt-and-pepper, has a white
shock, or is dyed). If the amount of hair is concluded to be one of the con-
cepts, the valid values for that concept must be determined.

Before continuing too far down this path, it would be wise to step back
and look at the original concept. A data element may exist in personnel