STANDARDIZED QUESTIONNAIRES 269

Older respondents were able to choose between the attributes in the first pair,
but could not understand what the other two pairs had to do with the way their
building looked.

Question construction has developed into a complex skill—perhaps too
complex. The semantic differential scale exhibits a problem many such tech-
niques face: They may cause more damage than they are worth. It is unclear, for
example, that seven-point polar—opposite judgment tests yield more information
than a three- or five-point agree/disagree rating scale. If respondents feel that
the adjectives they are asked to rate are nonsensical (cheerful and dreary applied
to a chair, for example), the loss of rapport with the interviewer may invalidate
other parts of the interview. Careful pretesting is one way to avoid such mis-
takes. Another is to include on the team constructing questions some people
who are similar to potential respondents. This is good advice no matter what
type of question you are constructing. In sum, rating scales of any sort must be
used only after careful examination of their wording and the operational
assumptions they embody.

0 Rank-ordering of items. It may be useful to pre-code responses to questions
asking respondents to rank a group of items relative to one another on a single
attribute such as importance, beauty, usefulness, or worthwhileness. For example:

Which of the spaces on the following list do you feel it is most important to
include in a house? (Please circle “1” for the most important, “2” for the second
most important, “3” for the next, and so on until you have ranked all places in
terms of their importance to you.)

bathroom 1 2 3 4 5 6 7 8 9 10 l 1 12

kitchen 1 2 3 4 5 6 7 8 9 10 l 1 12

laundry room 1 2 3 4 5 6 7 8 9 10 l l 12

living room 1 2 3 4 5 6 7 8 9 10 1 1 12

bedrooms 1 2 3 4 5 6 7 8 9 10 l l 12

den or rec. room 1 2 3 4 5 6 7 8 9 10 1 l 12

study 1 2 3 4 5 6 7 8 9 10 l l 12

storage attic l 2 3 4 5 6 7 8 9 10 1 1 12

vestibule 1 2 3 4 5 6 7 8 9 10 l l 12

dining room 1 2 3 4 5 6 7 8 9 10 l 1 12

Other: specify 1 2 3 4 5 6 7 8 9 10 1 1 12
Other: specify 1 2 3 4 5 6 7 8 9 10 l l 12
Other: specify 1 2 3 4 5 6 7 8 9 10 1 l 12

When items in the group are unduly complex—alternative lifestyles, for exam-
ple-it is easier to present them in pairs for sequential comparison rather than in
a simultaneous list.

Each technique for pre-coding responses creates opportunities for
researchers, but also limits what researchers can do with the data. Only experi-
ence with using classification and scaling methods—asking the questions, record-
ing answers, tabulating responses, and analyzing data—will give researchers the
knowledge and self-confidence needed to choose a form for pre-coded responses.