Information Technology Reference
In-Depth Information
systematic way of identifying studies for investigation. The approach used was based primarily on cita-
tions appearing in the ISI Web of Science. It is important to acknowledge, however, that the sample
could have been developed in a number of ways. Nonetheless, there is no reason to suspect that bias
was introduced in the approach used or that bias would have occurred using different approaches.
A related issue is the completeness of the sample of studies investigated. Because of time con-
straints it was not possible to examine all the studies that referenced the cornerstone cognitive fit
paper (Vessey, 1991). However, it does appear likely that, had we done so, we would have identi-
fied further studies that fit the profile sought here. Not only do indexing services work in specific
ways, so, too, do authors. It is possible that we did not identify studies because the authors did not
signal sufficiently strongly their theoretical bases.
As an example of the vagaries of identifying desired articles, we cite an article we became aware
of during our study, one that we do not examine here because it does not address the core issues.
This empirical article cited Vessey and Galletta (1991) twenty-two times, indicating that this paper
was very important to what the article was seeking to achieve. Yet the article did not acknowledge
the theory of cognitive fit as its theoretical base, with the result that the paper would not have been
identified in either an abstract or author-generated search for keywords. Further, Keywords Plus
would not identify the paper as being based on cognitive fit had we sought studies that referenced
Vessey and Galletta (1991) because keyword selection is based on “words and phrases that fre-
quently appear in the titles of the article's references,” and this paper cited just a single cognitive fit
paper. This example illustrates the difficulties associated with determining a complete sample of
relevant studies. Note, also, that the problems identified here present a problem for researchers who
believe that identifying core theories is key to the success of our discipline (see, for example,
Benbasat and Weber, 1996).
Another potential limitation lies in the classification system used to investigate the studies. As
noted earlier, we considered numerous classification systems, and discarded the majority of them
based on one of the three accepted tenets of such a classification system, plus our own fourth cri-
terion, that of balance in the resulting categories. We found that studies could be classified most
cleanly using the approach employed here. Yet a number of studies could have been classified
according to more than one of the characteristics used. For example, Speier and Morris (2003),
which we examined as an example of cognitive fit in more complex tasks because it illustrated
strategy change, used a multi-criteria decision making problem, as did those studies examined
under multi-criteria decision making in map-related domains. Nonetheless, the approach used
here presents an interesting and potentially fruitful way of viewing the potential of the theory of
cognitive fit.
Implications for Future Research
Our research has several implications for researchers.
First, to make the theory more explicit, it would be advantageous to be able to better define the
relationship between accuracy and time. As we have seen, the effects may be apparent in one or
the other of these dependent variables, depending on the emphasis placed on problem-solving
outcome, or they may be apparent in both. There are some pointers in the literature as to how this
issue could be addressed.
• Addo (1989), as presented in Wilson and Addo (1994), for example, presents a way of assessing
task complexity associated with “graphics problem solving.” Using this approach, Wilson and
Addo address the complexity of tasks in graphics problem solving as follows: (1) the number and
Search WWH ::
Custom Search