D. Scaling Method
All components were computed as composite
scores of five items using the five-point Likertscales
where 5 for strongly agree, 1 for strongly
disagree. A higher score indicates higher customer
retention.
E. Instrumental Bias
In administrating the questionnaire, there has been
a significant degree of bias that was controlled by
the researcher. The questionnaire was developed
based on the respondents. This minimized the
respondent bias to a certain level. Various scaling
methods were utilized where appropriate to
minimize bias.
F. Sample Size
In this study a two-stage random sampling method
was used. In the first stage 15 companies were
randomly selected from other companies in the
area. In the second stage, approximately 150
employees again were randomly selected from the
target population.