problem facing a researcher or manager by reducing the number of
dimensions that must be considered. In a recent paper, Kuhn ( 1987)
provides a systems-theoretic context for such a process. As Kuhn points out,
the strategic positioning process for a firm is both dynamic and relative; it
is a homeostatic process that, from the perspective of an observer, takes
place in a system composed of the firm, the competition, and the
environment of the firm.

The concept of strategic position implies that it is a firm's position
relative to a selected reference group or set of reference groups that is of
prime importance in determining strategic performance. Strategy should be
aimed at improving the position of the firm in the strategic space. Positional
emphasis would, for example, avoid the pain of attempting to improve on
historical performance in times when absolute increases are nearly
impossible. The emphasis would be on improving position vis a vis
competition; this might mean moderating a decline in absolute performance
to yield an improvement in ordinal position. Emphasis on affecting strategic
position would force the consideration of the possible actions and reactions
of other firms in the reference group and their effects on the resulting
position of the firm.

This book has introduced the concept and presented the basic
methodology of longitudinal analysis based on ordinal data. While several
articles have appeared in the academic and management literature, this book
represents the most complete introduction, development, and sets of
applications that is available for ordinal time series analysis. Management
researchers in general, and those interested in corporate strategy and policy
in particular, have traditionally eschewed the use of ordinal data in their
analyses. A reliance on cardinal data alone has resulted in a dearth of
longitudinal analyses--at a time when it has become widely recognized that
longitudinal analyses are necessary to further our understanding of strategy
and policy. Cardinal data have not been generally available in the quantity,
and especially in the quality, required for traditional time series analyses.
Nor is the situation likely to improve with the retrenchment of government
and its withdrawal from the collection and dissemination of time series data
regarding the private sector. The replacement of the government in data
collection and dissemination operations by private firms that must directly
cover the costs of their operations has meant that data quality and quantity
have improved in some cases, but that the cost of such data is often
prohibitive.

Even if the required cardinal data were available in the quantity and
quality desired, the necessity of employing sophisticated mathematical

-240-