The methodology of this research is based on Means-End Chain Theory. In this research, senior managers or company owners were interviewed to collect data by using soft laddering. Content analysis will be used to encode the respondents’ answers and catalogue into Attributes, Consequences and Values. Next, Implication Matrix is applied to analyze the relationship linkages between each factor and construct the Hierarchical Value Map which can visualize the result.
3-1 Target
The potential target companies were chosen from the register provided by the Taiwan Industrial Fasteners Institute and narrowed them down within the Kaohsiung area. Because GSCM certification is still not popularized in screw industry, this research will more focus on companies with ISO 9001 certification and their business partners. ISO 9001 is an international standard for QMS and GSCM certifications usually base on this standard to construct detailed procedures. Therefore, if a company have this certification, they may have much understanding of GSCM. Another resource of target companies is the media or business magazines, which can provide some companies already have had the green production line or green management.
The target companies were contacted by telephone and E-mail to confirm they were willing to interview or not. The qualification of respondents should be senior managers who had serviced this company for years or instinctively knew their supply chain management policy. Some obstruction may encounter during the process of connecting target companies and interviews. First, most companies are very concerned about commercial confidential issues; this research may encounter lots of rejection.
19
Second, recording equipment may be prohibited by respondents and it will increase the difficulty in collecting information. During the interview process, the researcher would mark the key phrase and confirm their meaning carefully to prevent information missing or misunderstanding. All the interviews were controlled within 30 to 45 minutes. After the interview, extra questions about other business partner’s information would be asked, which could improve the success rate of contacting the next company.
3-2 Laddering
This research used a semi-structured questionnaires (Appendix IV) and interview with soft laddering. According to Reyonlds and Gutman (1988), the respondent was informed that there would be no right or wrong answer, which allows participants to express their views in their own words and allows researchers to probe deeper into the responses. Three main questions are relative to A-C-V catalog. Before the beginning of the interview, a simple chit chat can help respondents who are more willing to discuss;
from previous personal experience, respondents usually cannot tell apart the different between Attributes, Consequences, and Values. Therefore, the interviewer has to provide practical example and connect to GSC issue. After a few times of questionnaire test, the best question sequence is C-V-A; “What are the possible consequences when you apply GSC ?”, “How will those consequences make you feel ?”, “What do you think about GSC ?” Due to the feature of semi-structured questionnaire, the interviewer can question respondents deeply and check the real meaning of answers.
3-3 Data Analysis
This research is constructed by MEC approach; the data analysis will process Content Analysis, Implication Matrix and Hierarchical Value Map. Kassarjian (1977)
20
mentioned that Content Analysis could extract the information from the interview context into quantitative data. This process includes information collection, context condensing, sampling, quantization, encoding and analysis (Krippendorff, 1980). First, translate the interview transcripts into proper phrase element and delete the redundant or unimportant information. Second, encode all phrase elements that are relevant for each level of the MEC (Reynolds, Dethloff & Westberg, 2001). During the coding process, the interview transcript includes many abstract phrases which may cause misunderstanding and coding errors. Therefore, a high degree of reliability can increase the overall consistency and intersubjectivity of the research. This research will invite 3 master degree students who are quite familiar with MEC process and 2 senior screw company’s managers to check the coding content. This research will use Stemler’s (2001) advice to test the Intercoder Reliability. The process includes two steps; first, three encoders should inspect the interview transcript and coding content. Second, using the formula for Intercoder Reliability = (2M)(N1+N2); M is the number of the elements agreed by Encoder 1 and Encoder 2. N1 means the number of variables agreed by Encoder 1; N2 is the number of variables agreed by Encoder 2. And the Reliability = (n Average Intercoder Reliability )[ 1+(n-1) Average Intercoder Reliability ]; n means the number of encoders. According to Kassarjain’s (1977) suggestion, the Reliability should be over 0.85 and then it can be considered an effective coding process.
Once finishing all the coding process, the elements will be put into Implication Matrix to calculate the frequency in each coding element. The Implication Matrix can conclude all linkage results which can be the basic data to develop the HMV (Reynolds
& Gutman, 1988). Each frequency represents the strength of the relationship, the higher frequency and the stronger the connection existing between each factor.
21
But if the HMV displays all the linkage form A-C-V relationship, the diagram will be too complex to show the significant connection. According to Gengler and Reynolds (1995) suggested, the cutoff value can be 5% of sample size; Reynolds and Gutman (1988) set the cutoff value from sample size’s 6% to 8.3%. When the frequency is over the cutoff value, it will be considered as effective value and displayed in the HVM.
22