Analysis
Since the survey instrument is composed of
factors taken from several studies, it is necessary to establish its internal reliability (Cronbach and Meehl, 1955; Hinkin, 1995; Peterson, 1994; Singh, 1995). To address this issue
Cronbach s alpha (α)was calculated for each
of the constructs and then for the entire
instrument (Peterson, 1994). In addition,
factor analysis was employed to examine if
there was any separation among the items
within the constructs that could not be predicted. According to Nunnally (1978), a
suggested cut-off value for α is 0.70. However, for exploratory research 0.50 is acceptable. In trying to obtain the best internal
reliability, those items which did not increase
α or did not match the remaining ones (items)
in the constructs were dropped. The initial
and final Cronbach’s alphas are listed in
Table II.
The government policy construct yielded
an α = 0.9005 when all the six items were
included (Table II). However, when the
co-operation with foreign investors was not
included the value increased to 0.9355. The
purpose of disregarding this item was twofold:
it did not seem relevant to inventory issues,
nor did it improve the internal validity. Contrary to expectations, the wage rate also did
not appear to fit the infrastructure construct.
This could be explained by the fact that the
labor content in the management of materials
is small. On removing it, the Cronbach’s
alpha increased from 0.8341 to 0.9150. The
product characteristics constructs yielded
α = 0.7091. No further changes in this