Control variables
Overall, the control variables have only a small impact on E-HRM user acceptance.
Differences between the two branches within Philips were, however, found to be
significant. This effect was not found at the management level, but only at the
shop-floor level. A direct theoretical interpretation as to why a commercial shop-floor
environment would be more conducive to a positive attitude towards E-HRM than a
shop-floor staff and support environment is not obvious. Local cultural differences
and/or local differences in shop-floor orientation towards management interventions
directed at cost efficiency might explain these findings, but this is speculation. No
significant effects were found for the other control variables (age, gender, knowledge of
IT, job experience). For knowledge of IT some degree of effects was expected based on
the literature. Venkatesh et al. (2003) include previous IT experience as one of the prime
moderators in their unified theory of user acceptance. The importance of this variable
was not confirmed in this study, but this may be caused by limitations in our
measurement of this variable.
We expected that user experiences would be very different between job types, but
this was only partially confirmed. Whereas HR literature places strong emphasis on
the differences between actors/stakeholders and between the levels in the organization
where these actors operate (Lepak and Snell, 1998; Rue¨l et al. (2004), the current study
does not confirm the necessity of such an emphasis, as far as HR roles are concerned.
On this issue the results for managers and shop-floor employees converged.
Differences were found, however, between management on the one hand and shop-floor
employees on the other hand concerning the technology acceptance variables. Whereas
usability appears to be an issue for managers and shop-floor employees alike,
experienced user support appears to be of special importance for managers.
Limitations
Prior to making recommendations based on these results, a number of comments have
to be made. First of all, the fact that the questionnaire data were collected on-line may
have caused selective non-response, because people with a negative attitude towards
IT and with less IT skills are more likely not to respond to an on-line questionnaire.
The relative overrepresentation of management and staff employees in the sample may
also have had a biasing effect. It remains to be seen if the findings, mainly those
relating to preferred HR roles, would be comparable if a more balanced sample as to job
type were available. Another limitation is the cross-sectional nature of the study. We
cannot draw any conclusions as to time precedence or causality from the findings.
Also, the preferred HR roles of strategic partner and change agent were not found as
separate factors in our analysis. The validity of the instrument used here to measure
HR role preferences needs more empirical confirmation. The measurement of the
“knowledge of IT” concept has been based on self-reports, and should be interpreted
with caution. Real knowledge and reported knowledge may differ. Moreover, this
concept is measured with a single item, which means its reliability is uncertain. We
also have to mention the forced omission of the concept “output quality” from the
analysis, owing to unreliability of the measurement scale. Finally, the specific setting
in which this study has been conducted is important. The results can probably be
generalized to other, central branches within large European organizations in the
industrial sector. Generalizing outside of this context, for instance to smaller
organizations, other sectors, more peripheral parts of a company, or other parts of the
world can only be done with a lot of caution.