index (e.g., formal grievance procedure and a policy of promotion from within)
may actually be part of a bureaucratic HRM system. In addition, the 75th
percentile cut-off is somewhat arbitrary. At this point it is unclear whether this
measure is appropriate for further use in SHRM research. At the very least
researchers should explore its assumptions prior to its use.
One final alternative is not to compute a single HRM system measure. For
instance, if it is believed that there are a small number of HRM practices or
policies that are the critical elements of the system, then researchers could use
individual measures of these practices and their interactions in regression
equations predicting performance. This approach, however, is still not without
concerns. To use this approach it would be important to have a very large
sample of firms. Using individual practices and their interactions uses up
many degrees of freedom and reduces the power to detect such interactions.
Additionally, there could be problems of multicolinearity among the HRM
practices and the interactions. Finally, higher order interactions may be impossible
to interpret. This technique may show promise, however, if the list of
practices can be narrowed down substantially and only a handful of theoretically
meaningful interactions are tested (e.g., Delery, Gupta, & Shaw 1998).
From this discussion it is clear that each of the methods used to compute a
measure of an HRM system is problematic. I believe that this is a critical issue
for SHRM research to explore and encourage future research that addresses
this issue.
Mediating Variables
The consensus among SHRM researchers is that HRM practices and systems
do not lead directly to firm performance. Rather they influence firm
resources, such as the human capital of the firm, or employee behaviors, and it
is these resources and behaviors that ultimately lead to performance (Wright
et al. 1994). This implicit model assumes that there are mediating variables
between HRM practices and firm performance, yet to date, few if any researchers
have measured these mediators or adequately addressed their importance.
It is important that future research attempt to specie the mediators and
attempt to collect measures of these constructs. This issue is crucial for continued
theoretical development in SHRM. By putting greater attention on these
variables, we may be able to better test how HRM practices influence these
constructs, and ultimately firm performance. In fact, focusing on these mediators
may help SHRM researchers identify systems of HRM practices that produce
them. Doty and Delery (1997) argued that HRM practices influence firm
performance by creating a work force that is skilled, motivated, and empowered.
Using this framework, researchers would focus attention on describing
how HRM practices influence these work force characteristics and how*
these characteristics can be measured. This should help in identifying and
measuring HRM systems.
Regardless of the specific framework, researchers need to attempt to mea-