For this empirical study, the dependent variables are each
relationship between the factors, and the independent variables are each practitioner that answered all the questions
from the survey. So, we have seventy-one dependent variables and forty independent variables. The survey has a
total of twenty-one questions and for each question we performed a hypothesis test. The goal was to identify if there
is any difference between the relative magnitudes of the relationships involved in each question. So, the null hypothe-
ses claim that there is no difference. For questions that
had more than two relationships, we performed Friedman
test. Otherwise, we performed Wilcoxon test. For the questions that we performed Friedman test, whenever the null
hypothesis was refuted, we performed Wilcoxon-Nemenyi-McDonald-Thompson post-hoc [13] test with the purpose of ordering the relationships by its relative magnitude. To
calculate the margin of error and confidence level given the
2sample size, we used SS = Z p(1−p) given in the work of
C2
Bartlett et al.[2], where SS is sample size, Z is the Z value,
p is the percentage picking of a choice and C is the confidence level. Given that the sample size is forty, the margin
of error calculated is 13% and the confidence level is 90%.
The survey results are presented as follows.