6.3. Pre-testing and validation of measures
All items and the survey instrument were pre-tested
with the help of 23 senior IS executives. Each of these
managers was actively involved in strategic IS
planning and each had significant experience in IS
management. All organizations were visited by one of
the researchers and face-to-face interviews conducted
with each manager. Assessments were made of the
items, constructs, and completeness of the instrument.
Some items were slightly refined and a preliminary
assessment indicated that there was a high degree of
internal consistency.
The complete set of measures for the six SISP
variables, the four effectiveness dimensions, and the
two contextual variables is described in Appendix A
with a description of the confirmatory procedures and
results of testing the psychometric properties. In
general, the high factor loadings and resulting strength
in measures of factor reliability suggested that each
scale exhibited strong characteristics of unidimensionality.
Further evidence of unidimensional measurement
was found in the significance of chi-square
values obtained in paired testing among constructs.
Items dropped due to large errors or equivocality were
noted. As shown, only single items from the scales of
flow, focus, and consistency were lost due to error.
Upon establishing the measurement efficacy, the
resultant factor loadings of each construct were used
to compute a factor score that represented a composite
measure free from random sources of error.