carrier matrix might differ in terms of the number of different sources from whom information is sought, and depth, in terms of the number of different messages involved. Diversity in the number of people contacted and the depth of each contact have both been found to be very important to successful performance within research and development laboratories ( Allen, 1977).
The results of the various tests of the CMIS suggest it represents a general framework for examining information seeking, but that further work needs to be done to incorporate contextual, and underlying factors in the model. This is a subject we will return to in Chapter 8. Most importantly, the various tests of the model have focused on vastly different information-seeking situations for respondents. Two of the prior research studies have focused on cancer- related information seeking by the general population ( Johnson, 1993b; Johnson & Meischke, 1993). Needless to say, for most individuals this is a nonrecurring problem, which is focused, novel, and fraught with emotional complications. These tests suggested that the model works best with authoritative channels, which are the object of intense, goal-directed searches, such as doctors.
The third study, which focused on information seeking in the defined context of a technical organization, yielded some critical differences in this more rational and programmed task ( Johnson et al., 1995). The most important variables in this test were those related to an individual's existing information base, those associated with an individual's need for recurrent, programmed information seeking, and those drawn from Johnson's model of Media Exposure and Appraisal. These results suggested that information seeking might be less influenced by motivations and individual cognitions than previously thought, and may be more correctly viewed as a beginning and end, in and of itself, done often to fulfill the requirements of someone's role in a social system.
In addition, as in prior tests of the MEA, only a minimal relationship was found between appraisal/utility variables and exposure ones, thus calling into question the current academic literature's fascination with processes of channel selection in organizations ( Rice, 1993; Sitkin, Sutcliffe, & Barrios-Choplin, 1992). Tests of the model suggest that barriers and various irrational processes may be more important than was initially thought, which is a problem we will turn to in more detail in Chapter 5. We will also return to the CMIS when we discuss future directions of information seeking in Chapter 8.
As we have seen, a whole host of factors cause the selection of an exponentially proliferating array of channels. The focus on channels is somewhat