19.4 Research and Evidence
on Internal Fit
.........................................................................................................................................................................................
Now that we have discussed the theory and complexities behind the internalWt
perspective, we turn to the empirical evidence. Until now, Delery’s ( 1998 ) review is
the only one that devoted considerable space to measurement issues and empirical
evidence of internalWt. Other review articles center around the strategy—HRM
system relationship or externalWt (e.g. Wright and Sherman 1999 )orWt in relation
to a speciWc activity area (e.g. compensation, Gerhart 2000 ). There is therefore
a need to describe and evaluate the empirical results related to the diVerent types of
internalWt. This section centers around inter-practice and intra-practice areaWt
since most of the research to date has investigated only these types of internalWt.
We could notWnd studies that speciWcally addressed within-HRM system vertical
Wt or between-HRM systemWt. The lack of studies in both of these areas is
troublesome.
- 1 Inter-HRM Activity Area Fit
As previously conceptualized, inter-HRM activity areaWt denotes theWtacross
diVerent HRM activities at the same level of abstraction (e.g. the alignment
between compensation and work design HRM practices). In aWrst step to inves-
tigate this type ofWt, several authors have examined whether certain HRM activ-
ities hang together and make up a theoretically coherent HRM system. In one of
the earliest studies following this line of thought, Arthur ( 1992 ) used a cluster
analysis approach and grouped organizations based on ten HRM policy and
practice variables that made up a HRM system. As predicted, Arthur found that
organizations with a cost leadership strategy had substantially diVerent HRM
systems from organizations with a diVerentiation strategy.
Empirical researchers seem to have followed Becker and Gerhart’s ( 1996 ) call for
more research using cluster analytic techniques (e.g. De Saa ́-Pe ́rez and Garcı ́a-
Falco ́n 2002 ; Delery et al. 1997 ; Ordiz-Fuertes and Ferna ́ndez-Sa ́nchez 2003 ; OstroV
2000 ). In general, these studies support the notion that there are separate HRM
systems, comprised of distinct HRM activities (mostly at the policy and practice
levels of abstraction). For example, OstroV( 2000 ) foundWve, De Saa ́-Pe ́rez and
Garcı ́a-Falco ́n( 2002 ) four, and Ordiz-Fuertes and Ferna ́ndez-Sa ́nchez ( 2003 )two
clusters of distinct HRM systems. There are likely to be many reasons for the
diVerent numbers of clusters found, ranging from sample characteristics, HRM
activities included in the survey, a mixture of policyandpractice items, through to
shortcomings of cluster analysis, which are well documented (see Delery 1998 ).
396 sven kepes and john e. delery