satisfaction for its asymmetric and nonlinear nature. This is to proceed quality classification. This paper has opinion to intemperate word description into statistic language in must and attractive quality attribution in Kano’s two dimensional model.
It is to combine statistic and graph technique to replace the original Kano’s evaluation table to approve the practical of this paper. This paper will take banking service as example.
Section 2 The New Kano Categories Way by Piecewise Regression
When analyzing a relationship between a response, y, and an explanatory variable, x, it may be apparent that for different ranges of x, different linear
relationships occur. In this study, a single linear model may not provide an adequate description and a nonlinear model may not be appropriate either. Piecewise linear regression is a form of regression that allows multiple linear models to be fit to the data for different ranges of x. Breakpoints are the values of x where the slope of the linear function changes .The value of the breakpoint may or may not be known before the analysis, but typically it is unknown and must be estimated. The regression function at the breakpoint may be discontinuous, but a model can be written in such a way that the function is continuous at all points including the breakpoints. To be specific, we have to find a piecewise linear approximation of the relationship between the attribute-level performance and the customer overall satisfaction. Given what is understood about the customer satisfaction, we assume the function should be continuous. In order for the regression function to be continuous at the breakpoint, the three equations for Y need to be equal at the breakpoint.
Let X be a polyhedron in the n-dimensional space Rn and Xi, i = 1,…, s, a polyhedral partition of X, i.e. Xi∩Xj = for every i, j, = 1,…,s and UiS1Xi = X. The target of a Piecewise Linear Regression (PLR) problem is to reconstruct an unknown function f : X→R having a linear behavior in each region Xi.
f(x) = zi = wi0 +
n
j j ijx w
1
(7)
Three straight lines (continuous):
E(y) = 0+1x1+2(x1-k2) x2+3(x1-k1) x3 (8)
Where
k1 and k2 are knot values, k1 < k2
0 if not k if
1 1 1
2
x x
0 if n o t k if
1 1 2
3
x x
To make things easier only three pieces in the concave piecewise linear approximate the curve. We impose the concavity restriction that is not present in
other works; moreover, and we want to solve this problem in SAS. This study conduct measuring scale from 1 to 10. Y be the customer satisfaction, X be the quality performance, and 4 and 7 be the knots, respectively. Then, based on promotional data, we need to find the set of unknown parameters B0, B1, B2, B3, that minimize the objective function (the sum of squared residuals):
10 X 7 for ,
) 7 ( 3
4
7 X 4 for ,
) 4 ( 4
4 X 1 for ,
j 22
1 3 2
1 0
j 22
1 2 1
0
j 22
1 1 0
j
j j j
j j
j
j j j
j j
j j j
X B B
B B
Y
X B B
B Y
X B B
Y
(9)
Mathematically, the problem of finding the curve Y, formulated above, is that of nonlinear programming with a continuous but non-differentiable objective function and linear and nonlinear constraints. This type of problem could be solved by SAS/OR Non-Linear Programming procedure (PROC NLP). PROC NLP has 11 different optimization algorithms. However, we can use only one of them, namely Nelder-Mead Simplex Method (NMSIMP). It is the only algorithm that doesn’t require first and second order derivatives, and allows for boundary constraints, linear constraints and nonlinear constraints.
Section 3 Case Study
I will take banking service as case study to test the method mentioned above meet the overall conceptual work of Kano’s model.
1 Method
I aims to focus on non-linear relationship of banking service quality performance and customer satisfaction which refers to the questionnaire content of banking service quality by Walfried and Chris (1998). The front 22 items are performance measurement of each quality factor and the 23 item is to investigate overall satisfaction of customer toward banking service. All items were modified in accordance with the nature of the industry. All statements were formed positively in a ten-point. The study object of this case study randomly sampled personal banking financial customer in Taiwan area. This paper conducts SPSS 13.0 and SAS/OR to
undergo credity and validity analysis. There were 850 copies of questionnaire released and 630 copies were collected as valid questionnaire. The validate feedback rate is 74%. Reliability was tested using the Cronbach’s α. This paper firstly put each quality factor performance (22 varibles) into non-linear planning function 9 to produce 88 variables. The selected reserve variable will be determined according to parameter calculation result and then it will be place into function 9 for output.
2 Research Result
Table 12 gives the Cronbach’s α for each of these dimensions. It also shows the factor loading of each item and its original dimension. The results shown in Table 12 indicate that five dimensions are reliable (Nunally, 1978). This supports the internal cohesiveness of the items forming each dimension. Moreover, the loading of quality factor under each perspective are all more than 0.7 which indicating it creditability.
This paper also revises the question item of Walfried and Chris’s questionnaire system while content validity was reviewed by expertise through pilot test so there should be no concern of validity. Table 12 also shows the mean, standard deviation for each of the variables. This table indicates that customers are generally happy with overall service quality and all the dimension of the instrument. In each attribute-level performance, “Zero service defect” has the worst performance while “Neat and tidy appearance of staff” has the best performance.
Table 12
Performance of each quality factor
Quality Elements factor
loafing Cronbach’s α Mean S.D.
Tangible Q1~Q4 0.745
1. Modernlizd facility 0.823 6.967 2.188
2. Attractive appearance 0.788 6.009 2.160
3. Neat and tidy appearance of staff 0.844 7.267 2.014
4. Overall coordination 0.851 6.730 2.210
Reliability Q5~Q9 0.768
5. Completing service in time 0.777 6.948 2.215
Table 12 (Continue)
Quality Elements factor
loafing Cronbach’s α Mean S.D.
6. Enthusiasm to solve customer’s
problem 0.762 6.776 2.326
7. Offering needed service at first
moment 0.857 6.506 2.410
Table 12 (Continue)
Quality Elements factor
loafing Cronbach’s α Mean S.D.
8. Commitment to follow up contract 0.719 6.021 3.032
9. Zero service defect 0.756 5.509 2.696
Response Q10~Q13 0.899
10. Service notification 0.874 6.379 2.539
11. Offering appropriate service 0.721 6.809 2.159
12. Willingness to help 0.707 6.948 2.279
13. No ignorance cause of busy 0.812 6.061 2.424
Assurance Q14~Q17 0.811
14. Customer has confident on the
bank because of its performance 0.895 6.639 2.102
15. Customer has secured feeling in
trade 0.786 6.994 2.099
16. Staff always be polite 0.805 6.891 2.264
17. Responding to customer with
professional knowledge 0.769 6.797 2.263
Affectiveness Q18~Q22 0.784
18. Offering personalized service 0.756 6.467 2.606
19. Convenience in service hour 0.787 6.415 2.338
20. Look after every customer 0.782 6.185 2.398
21. Customer’s benefit is priority 0.716 5.806 2.573 22. Understanding of customer’s
need 0.894 6.258 2.415
Using the results by the piecewise regression, we could draw these nonlinear curves and to observe the graphs for categorization. We demonstrate the use of piecewise regression as a statistical technique to model the relationship between the attribute-level performance and overall satisfaction. As mentioned in study methodology, we keep going the first output of the parameter calculation are total seven remarkable variables (those are quesionnaire items No. 2, 3, 4, 7, 11, 18, and 22); moreover, we substitute those variables into the nonlibear regression function to find out the parameter result. The output of piecewise regression parameter is showed as Table 13, among B1、B2 and B3 are separately represented the slope of piecewise regression curves at the scale 1-4、4-7 and 7-10 on the attribute-level performance. All piecewise regression curves are drawed on the Figure 6.
Table 13
Output of piecewise regression parameter
Quality Elements B0 B1 B2 B3
Attractive appearance 10.053 0.729 0.146 0.247286
Neat and tidy appearance of staff 4.672 -0.045 0.253 1.734571
Overall coordination -4.721 .02945 0.179 0.104571
Offering appropriate service -21.492 5.04725 0.593 1.450571 Offering needed service at first moment 0.473 1.277 -0.159 -0.24786 Offering personalized service -17.059 4.51825 0 0.842857 Understanding of customer’s need -8.036 1.68525 1.518 0.640571
Figure 6 Output of piecewise regression of each quality factor -10
-5 0 5 10
1 4 7 10
PERFORMANCE
Attractive appearance
Neat and tidy appearance of staff Overall coordination
Offering appropriate service
Offering needed service at first moment Offering personalized service
Understanding of customer’s need SATISFACTION / DISSATISFACTION
As shown in Figure 6, each quality factor has different piecewise regression curve. The increase or decline of parameter (different satisfaction caused by different quality performance) in each quality character is to meet the original conceptual work of Kano’s model for classification. It is to say parameter of each quality character is must-be quality which is B3j < B2j < B1j. On contrary, B3j > B2j > B1j, represents as attractive quality. Table 14 shows that the result of Kano attribute classification in each quality character is applied by piecewise regression. In the 7 items of quality characteristic, there are 5 must-be factors and one in each of attractive quality factor and one-dimensional quality attribute.
Table 14
Result of Kano attribution classification
Quality attribute Characteristic of Banking Quality Attractive Quality Neat and tidy appearance of staff One-dimensional Quality Attractive appearance
Must-be Quality
Overall coordination
Offering needed service at first moment Offering appropriate service
Offering personalized service Understanding of customer’s need
3 Conclusion and Suggestion
Due to the reaching climax of banking competition, the case study shows that the implicit of bank has less difference services which does not stimulate or encourage consumers (Must-be quality). Customer orientation is the surviving tip for banks. The banking service has been standardized, such as electron fund convertible system, since the implementation of Taiwan financing service license and the awareness of professional ability build-up from management level in bank. The offered services include innovation of coping with customer waiting line.(ex. set up specialized counter or wait with number ticker) as well as upgrading of financing organization flexibility and immediate information processing ability. However, there is lack of addressing on service ambience for customer to experiencing the vision environment and atmosphere (ex. sense of beauty). Atmosphere arranging is to stimulate customer recognition and behavior to enhance customer satisfaction. The priority financing habit is that customers prefer to visit bank themselves. The result shows that there is less performance difference in ATM service or internet banking in
each bank when stuff does not offer service. Hence, customer is used to evaluate banking service through their feeling at the scene when referring to banking service evaluation. It is recommended that banks to emphasis on atmosphere creation by applying systematic product or mechanical method and psychology as supporting instrument on the purpose of building up quality difference to be competitive. The previous works of this area has applied Kano’s model to discuss bank service, such as Bhattacharyya and Rahman (2004) applied the Kano’s model to understand what drives customers’ satisfaction and dissatisfaction in a bank in India. We conducted pair wise questionnaire and 5×5 evaluation table to proceed Kano’s two dimensional survey in 2004. To response the transform of time and banking management environment, the previous survey result is differing from this time. Partial attractive quality declined as clear orbit of one-dimensional or must-be quality (Lee, 2004). By identifying attributes that should be emphasized or de-emphasized, Kano’s model guides the development of action plans to minimize mismatches between satisfaction and attribute-level performance, resulting in improved operational efficiencies through resource redeployment recommendation.
This paper investigated the service quality categories based on the Kano’s model. A questionnaire based on the original instrument of SERVPERF and Walfried and Chris (1998) was developed. This questionnaire was distributed to 850 customers in Taiwan and 630 responses were received. These factors were all reliable. This paper attempts to solve the non-linear problem between attribute-level performance and the overall satisfaction by formulating piecewise regression whose solution presumably gives a near-optimal solution to the original problem. The piecewise regression model, as applied here, is a function whose definition of the relation between the attribute level performance (X) and customer satisfaction (Y) depends on the value of the input. Therefore, piecewise regression could proceed simplified non-linear question through a different equation if each phase characterized. This study is based on major quality classification statement of Kano’s model to propose a new method combining statistic and graphing technique to replace the Kano’s evaluation table. It is to apply piecewise regression to graph the relationship curve of quality performance and customer satisfaction in different quality attribution. It is with attempt to reveal the actual quality attribution.