Equation 3. 1 Computation of Sample Size Source: (Yamane, 1967)
3.5 Data Collection Methods 1 Type and nature of data
Both primary and secondary data were obtained and utilized for purposes of addressing the research objectives. Secondary data was sourced from both published and unpublished records such as the National Environmental Management Authority, Kenya Association of Manufacturer and Ministry of Environment annual reports, journals and books. Primary data was collected on GSCM practices (green procurement, green manufacturing, green distribution and environmentally-oriented reverse logistics), supply chain ecocentricity (moderating variable) and also on firm performance indicators as shown in Table 2.2.
3.5.2 Data collection instrument
The main primary data collection instrument was a survey questionnaire consisting of structured closed and open-ended questions (see appendix I). The questionnaire consisted of six key parts, all aimed towards capturing the relevant information in respect of the study objectives. Part 1 dealt with general information about the respondent and the firm, Part 2 facilitated capturing of data on various green SCM practices adopted by firms, Part 3 facilitated recording of data on the moderating variable (supply chain ecocentricity), Part 4 dealt with performance of firms, Part 5 facilitated capturing of data on the realization of green SCM practices benefits, lastly, Part 6 dealt with the number of years of practicing GSCM by the manufacturing firms. The primary data was captured using multiple choices questions and a five point likert type scale. In applied management studies, the likert type scale is one of the acceptable techniques for measurement of attitudes in a ―scientific‖ way which allows the use of statistical tools to analyze data (Blaikie, 2003).
3.5.3 Instrument Administration
The study used an e-mail survey to collect primary data to test the hypotheses generated in Chapter 2. E-mail surveys are employed extensively in research due to their ease of use, flexibility of responding, confidentiality and relatively low-cost (Dillman, 2000). Online surveys are easily quantifiable and suitable for statistical testing, as the results are
typically collected in a file that is easily manipulated for analysis. In addition, e-mail surveys reduce the degree of interviewer bias and are appropriate for collecting a large number of geographically dispersed respondents in a cost-effective manner (Dillman, 2000)
The challenge of e-mail survey in this study was gaining the trust of potential respondents. With the deluge of e-mail traffic that most business professionals receive, potential respondents were reluctant in taking part in the survey, believing it to be an internet marketing promotion. A second challenge was of a survey methodology in general. Researchers often find that business professionals do not have time to complete a survey and/or are over-surveyed, resulting in ―survey fatigue‖ (Cooper & Schindler, 2006). These challenges were addressed through employing a two-phased approach to reaching potential participants. The first phase consisted of sending out a mass e-mail to a list of potential participants using the outlook e-mail program. Outlook allowed sending of individualized e-mails to each potential participant containing a reminder alarm of a completion date of two weeks from the send date of the e-mail. The program was automatically activated to send reminder alarm at the end of the two weeks window to the respondents. However, a polite reminder e-mail was sent to those who may have not responded at the end of the two weeks window.
The second phase started after the final reminder e-mail. Once the results of the personalized e-mails and reminders were collected, the remaining valid contact information for participants that had not responded to the survey, a follow-up contact was made as reminders until the completed survey was received. Follow-up contact included additional reminder phone calls and e-mails.
3.5.4 Data Retrieval and Response Rate
The returned questionnaires were checked for consistency and validity of the respondents‘ answers. An effort was further made to control the research process through the installation of anti-sperm software on the outlook system to prevent e-mails from un- recognized sources from finding their way into the questionnaire file. A random double
checking for authenticity of the data was conducted to ensure the officers who were the unit of observations did not delegate the filling of the questionnaires to their assistants. This was however established to be not an issue in this study. To improve the response rate, a continuous follow up were made by phones and e-emails after the initial e-mail contact which resulted in receiving back a total of 179 out of 234 responses. However, 18 were found to be incomplete and therefore were not analyzed. This left 161 valid responses or 69 percent which is considered high rate of returns for a survey research (Keeter, Scott, Kennedy, Dimock, Best & Craighill. 2006). According to Richardson (2005), 50 percent response rate is regarded as an acceptable in a social research survey. Baruch (2007) established that the average response rate in social research surveys is 55.6 percent. Therefore, the study valid response rate of 69 percent was considered high and acceptable for this study.
The study is a wake to the fact that data is ordinarily received in different forms. Cooper and Schindler (2006) suggested two of the formats to be textual and numeric data. This study collected both for addressing the study objectives. Malhotra (2004) explains that data preparation precedes data analysis. The process of data preparation imparts on data accuracy and enforces a conversion from raw to classified data that can benefit analysis and interpretation. Therefore the study applied coding, editing and tabulation as forms of data preparation (Malhotra, 2004).