webster & watson - How to write a LR - MISQ How To Write A Literature Review | Page 7

Webster & Watson/Guest Editorial

Theoretical Development in Your Article
A review should identify critical knowledge gaps and thus motivate researchers to close this breach. That
is, writing a review not only requires an examination of past research, but means making a chart for future
research. For example, the MISQ Review articles by Alavi and Leidner (2001) and Te'eni (2001) pinpoint
questions for future inquiry.
Highlighting the discrepancy between what we know and what we need to know alerts other scholars to
opportunities for a key contribution. Usually, this roadmap is accomplished by developing a conceptual
model with supporting propositions. In this paper, we focus on this traditional approach. However, there
are other means of making a significant contribution (Whetten 1989). For instance, showing how competing
theories or philosophical assumptions explain an important phenomenon can be very influential (e.g.,
Allison’s [1969] analysis of the Cuban missile crisis).
Extending current theories or developing new theories will create directions for future research. However,
extending or developing theories is a difficult task and is often the weakest part of a review. Nonetheless,
it is the most important part of a review and generally needs the most elaboration. Here, we provide some
recommendations for researchers who wish to develop a model and justify its propositions.
Conceptual models are generally derived from variance (factor) or process theories (Markus and Robey
1988; Mohr 1982). Variance theories incorporate independent variables that cause variation in dependent
variables. In contrast, process theories use events and states to help explain dynamic phenomena. Thus,
models may look very different in the two approaches (see Figure 1 of Langley [1999] for examples of
these two types of models). Of course, review articles may draw from both variance and process research
to develop conceptual models to guide future research. In fact, DiMaggio (1995, p. 392) argued that “many
of the best theories are hybrids, combining the best qualities” of these approaches. Moreover, Sabherwal
and Robey (1995) demonstrate how the two approaches can be reconciled in one study. Thus, do not treat
the results of variance and process research as independent elements of a review. Rather, make every
effort to show how these two approaches reveal a deeper understanding of the topic. For example, the
explanatory power of a process model might be contingent on the strength of a particular variable, as
Newman and Sabherwal (1991) illustrate.
Models and propositions capture relationships between variables, but do not, on their own, represent
theory (Sutton and Staw 1995). For example, Griffith’s (1999) proposition 1a (p. 480) states that,
“New/adapted concrete features are more likely to be experienced as novel than new/adapted abstract
features.” Rather, the reasoning or justification for these relationships represents the crucial part of the
theory-development process.
The reasoning for propositions may come from three main sources: theoretical explanations for “why,” past
empirical findings, and practice or experience. The why or logical reasoning is the most important component of the explanation. It must always be part of any justification. It represents “the theoretical glue that
welds the model together” (Whetten 1989, p. 491). Past empirical research also should be included if it
exists. If it does not exist in the specific area of interest, however, empirical research in related areas
should be presented as (weaker) support (Gay and Diehl 1992). Experience, if available, can also help to
justify a proposition; it may arise from the author’s own experiences in interacting with organizations or from
the practice literature. Nonetheless, while past findings and experience can help to support a proposition,
keep in mind that they are not a substitute for logical reasoning (Sutton and Staw 1995).
As justifying propositions often represents one of the most challenging aspects of a review paper, let us
look at several examples. First, if we examine Griffith’s research concerning technology features, we see

MIS Quarterly Vol. 26 No. 2/June 2002

xix