266 CHAPTER II

Nominal codes are most useful to collect information, to offer non-
ranked choices to respondents, and to obtain attitudinal data useful in a bina-
ry “yes” or “no” form.

Ordinal

To analyze intensity, direction, and quality of such variables as verbally
expressed attitudes and perceptions, it may be helpful to arrange responses in a
rank order representing different degrees or magnitudes.

When each category is separated from others by what seems to be an
equal magnitude, ordinal categories are called intervals. There are some prob-
lems with this idea. For example, a uniform difference in temperature—say, 2°
F— may be experienced differently if it represents a rise from 6° F to 8° F than
a rise from 65° F to 67° F. The same problem holds for age differences. A year
has a different quantity and quality at ages five, thirty, and eighty. Therefore,
interval categories are not presented in the following discussion as distinct from
ordinal ones.

0. Information. Ordinal pre—coding can be used for questions gathering
information that are reasonably seen as “how much” or “how many” questions
regarding age, income, size of household, or the number of clubs a respondent
belongs to.

Age: Under 11 CI, 11—20 CI, 21—30 CI, 31—40 C], 41—50 CI, 51—60 CI, 61 or over CI.
Club membership: None El, 1 or 2 Cl, 3 to 6 Cl, 7 or more CI.

0 Attitudes. Ordinal coding may also be useful for response categories fol-
lowing questions that ask respondents to judge the intensity of an attitude
about a situation, person, object, or setting.

Would you say the rules in this factory are very fair Cl, fair Cl, unfair D, very unfair CI,
or do you have no opinion CI about this?

Would you say the work areas you have are very supportive CI, supportive Cl,
unsupportive Cl, very unsupportive D, or are you uncertain about this El?

When a questionnaire is administered orally, the “no opinion” or “uncer-
tain” category is sometimes not read to respondents, encouraging them to make
some kind of choice—no matter how weakly felt. If they still have no opinion,
interviewers check the box.

Some coding categories are associated with both a format for responses
and quantitative procedures for analyzing responses. One of these is the Likert
attitude scale, in which groups of statements are presented to respondents for
them to indicate the intensity of their agreement or disagreement. If standard
scores are assigned to responses in such a way that high agreement with posi-
tive statements is equivalent to high disagreement with negative statements, and
if several questions tap dimensions of the same general attitude (“feelings about
company management”), then cumulative scores on these statements can be
used to indicate a respondent’s position on that attitude.