To measure normative beliefs in a way that maximizes their ability
to predict subjective norms (and behavioral intentions and
behaviors), it is necessary to conform to the principle
of correspondence. According to this principle, behaviors
have four components. These are action, target, time, and
context. For example, consider the behavior of attending a
cervical cancer screening. The action is "attend," the target
is "cervical cancer screening," the time is whenever the screening
will take place (e.g., 2:00 next Tuesday), and the context
might be "at the local health center." The key measurement
point is that all of the variables one wishes to use must
be measured so that all four components-action, target, time,
and context-are precisely the same for the measures of all
of the variables. Even a small deviation from perfectly correspondent
measures can result in a dramatic decrease in the correlations
between variables (Davidson & Jaccard, 1975; 1979).
Suppose a researcher wishes to use normative beliefs to predict whether people will perform the behavior of "Attending a cervical cancer screening next Tuesday at 2:00 at the local health center." In addition, suppose the researcher has performed an elicitation study and has determined that "my doctor" is a relevant important other for the population of concern. The normative belief pertaining to "my doctor" could be measured as follows: "My doctor thinks I should (should not) attend a cervical cancer screening next Tuesday at 2:00 at the local health center." Participants would respond on a 7-point scale (ranging from -3 for an extreme score indicating "should not" to +3 for an extreme score indicating "should") that describes the extent to which the person believes the doctor thinks he or she should or should not perform the behavior. For the purposes of evaluating internal reliability, it may also be worthwhile to have additional similar items. Two examples might be as follows: "My doctor thinks it would (would not) be a good idea for me to….." or "My doctor would want (not want) me to …." These items should also be used with 7-point scales. Appendix A provides examples of items for measuring subjective norms, normative beliefs, and motivations to comply. Appendix B provides details about how to use the items to obtain summary scores for the variables and for using these summary scores for prediction and intervention.
Researchers sometimes cannot narrow behaviors down to a specific time and context. In the cervical cancer case, it may be sufficient that people get a screening "anytime in the next 6 months," and the place where the exam is obtained may be unimportant. In this case, the normative belief pertaining to "my doctor" could be measured as follows: "My doctor thinks I should (should not) attend a cervical cancer screening any time in the next 6 months." But a cautionary note must be sounded here. Because the context is unspecified in the normative belief measure, it must similarly be unspecified in the subjective norm measure and, ultimately, in the behavior measure. The failure to have corresponding measures of the different variables will be likely to result in low correlations between them. This means, for example, that measuring behavior by going to a particular cancer screening center to assess who attended or did not attend the screenings that took place there during the relevant 6 month period is insufficient because participants might have obtained their screening someplace else.