Page 99 - 33-3

NTU Management Review Vol. 33 No. 3 Dec. 2023

5
judgment matrices, we compute the consistency ratio (CR) from the matrices to examine
the response consistency and ensure that the validity in the respondent conceptualizations
of each goal increased with the lines of the goal description (e.g., “Goal A is more
important than Goal B” should be invariant as lines of detail are added to the description

of goals). Following Saaty (1977), the threshold value of the CR must be CR ≤ 0.10, but
a ratio less than 0.20 is considered tolerable. The results show that the average CRs at a
significance level of 0.05 for the group of regional managers and branch managers are

0.17 and 0.19, respectively. This indicates a moderate degree of consistency of judgment
among the respondents, indicating that the validity of our survey data, and the AHP is not
a significant concern.
After examining the consistency, we follow prior studies (Dong, Zhang, Hong, and
Xu, 2010; Wu and Xu, 2012; Chiclana, Mata, Martinez, Herrera-Viedma, and Alonso,

2008) and apply an AHP consensus model to measure the consensus index between
6
a regional manager and a branch manager. This number thus represents the degree of
consensus of the judgment matrix between a regional and branch manager and serves as

our CONSENSUS score for a regional and branch manager pair.
3.3.2 Branch Office Prior Performance (PERF )
i,t-1
To measure a branch office’s prior performance, we use an indicator variable
PERF , which was equal to one if the branch office’s (t-1)-th month of sales revenues is
i,t-1
ranked in the top 25 percent of the sales revenue within the same region and same month

and zero otherwise.

3.4 Measurement of Dependent Variables
3.4.1 Regional Manager Resource Allocation Preferences (Ab_SENIOR and Ab_ADV )
i,t
i,t
A regional manager has the right to distribute the amount of advertising funding
and to (re)deploy and adjust the number of salespersons in branch offices by way of

5 Unlike studies (e.g., Bedford, 2015; Guenther and Heinicke, 2019) that use rating scales to construct
their main variables and apply statistical methods such as Harman’s single factor score or three-phase
confirmatory factor analysis to examine common method biases in their survey data (e.g., Podsakoff,
MacKenzie, and Podsakoff, 2003; Podsakoff, MacKenzie, and Podsakoff, 2012), the AHP approach
uses the consistency ratio (CR) to measure data validity.
6 The consensus models are explained in Appendix B.

91