Equation 3. 1 Computation of Sample Size Source: (Yamane, 1967)
3.6 Pilot Study
To ascertain reliability, validity and reduce measurement error, a pilot test was conducted (Dillman, 2000). The objective of the pretest was to establish any potential problems with the design and instrumentation and to provide proxy data for selection of a probability sample (Billé, 2010). The researcher used the pretest to assess the clarity, complexity and the face validity of the measure. As a result, revisions were made that improved the overall look and content of the final data collection instrument in terms of readability, wording and arrangement (Teijlingen& Hundley, 2001)
A total of 20 respondents (manufacturing firms) were used in the pretest as recommended by Monette, Sullivan and DeJong, (2002) for a survey study. The respondents were drawn from the same population frame that was similar to the one used for the actual survey in terms of background features and familiarity with the study topic. The reactions received were instrumental in refining the questionnaire before it was finalized for the study. Background information obtained through pretesting process provided insights into the simplification and strengthening of the process in this regard, and allowed for greater understanding of the specific context and the respondents as individuals to the extent that the process was tailored to the specific context.
3.6.1 Scale Construction
The questionnaire was abridged after pretesting to attain a balance between data required and the time needed to collect the data and to decrease the chance of lethargy for the respondents. The final questionnaire was profoundly composed of simple and un- ambiguous closed and open ended questions designed for multiple linear regression analysis.
3.6.2 Reliability
The measurement of human behavior belongs to the widely accepted positivist view, or empirical analytic approach, to discern reality (Smallbone & Quinton, 2004). Because most behavioral research takes place within this paradigm, measurement instruments must be reliable. Reliability refers to the consistency of measurement (Bollen, 1989), or stability of measurement over a variety of conditions in which basically the same results should be obtained (Ritter, 2010). According to Smallbone and Quinton, (2004), a reliable measure is characterized by stability over time, and internal consistency. A measure would exhibit stability if little variation over time was found when the measure was re-administered and would exhibit internal consistency if the ―indicators that make up the scale‖ are dependable (Shadish, Cook, and Campbell, 2001).
The most popular method of testing for internal consistency in the behavioural sciences is coefficient alpha. Coefficient alpha was popularised by Cronbach (1951), who recognized its general usefulness. As a result, it is often referred to as Cronbach‘s alpha. According
to Smallbone and Quinton, (2004), the coefficient alpha is suitable in measuring variance attributable to the subject and variance attributable to the interaction between subjects and items. Zikmund, (2003); Ritter, (2010) provided a measuring scale for acceptable alpha as 0.60 and above. According to them, above 0.60 is considered as an indicator of a good internal reliability. This study therefore used the Cronbach‘s alpha to test the internal reliability of the measures. For the specific tests of internal reliability for the dimensions of GSCM, the Cronbach‘s alpha results are presented in section 4.3.
3.6.3 Validity
According to Knapp (1998); Carter and Porter (2000); Peat (2002), validity is defined as the extent to which the instrument measures what it purports to measure. It is a direct check on how the instrument fulfills its function. A test of validity is therefore whether the measure of a concept really measures that concept (Peat, 2002). There are several measures of validity that provide evidence of the quality of a study. Internal and external validity relate to the overall study design. Internal validity relates to the extent to which the design of a research study is a good test of the hypothesis or is appropriate for the research question (Carter & Porter, 2000). External validity, meanwhile, relates to whether or not research findings can be generalized beyond the immediate study sample and setting (Carter & Porter, 2000). Therefore this study used Peat (2002) measures to assess the validity of data collection tool as follows:
(a) Content validity
Content validity is a qualitative type of validity where the domain of the concept is made clear and the analyst judges whether the measures fully represent the domain. It is whether a tool appears to others to be measuring what it says it does. Face validity is a simple form of content validity. The researcher built content validity into the measures through the derivation of the scales from theories related to GSCM practices and firm performance (Carter & Porter, 2000) and also by incorporating comments from experts in GSCM in the content of the instrument (Peat, 2002). As such, the study considered the
content of the instrument to be valid implying that the study instrument measured what it was supposed to measure i.e. the measure fully represented the domain of the study.
(b) Criterion validity
Concurrent or predictive validity are both measures of criterion validity (Trochim, 2006). Concurrent validity uses an already existing and well-accepted measure against which the new measure can be compared. Predictive validity refers to the ability of a test to predict an event in the future (Smallbone & Quinton, 2004). Criterion validation therefore refers to the effectiveness of a measure in terms of being able to predict an event related to relevant criteria (Trochim, 2006). This was not however considered to be an issue with respect to the surveying of the respondents in the study.
(c) Construct validity
Construct validity is the degree to which an instrument measures the trait or theoretical construct that it is intended to measure (Shadish, Cook & Campbell, 2001). Construct validity was ensured in this study through derivation of the measures from GSCM theories which were to be tested in the study. To confirm construct validity of the measures, factor analysis was performed and the results of Eigen values is presented in Table 3.5
Table 3. 5 Result of Eigen values
Exploratory Factor Analysis (validity)
Variables Question Number of Eigen
Values of 1 or higher
Total variance explained
h2
Interpretation
Green procurement 1 63% Unidimensional
Green manufacturing 1 62.52% Unidimensional
Green distribution 1 70.32% Unidimensional
Environmentally oriented reverse logistics
2 79.14% Unidimensional
Eigen values are used to establish the construct validity of the instrument (Brown, 2001). According to Hair, Andrson, Tatham & Black. (2006) a factor with an Eigenvalue of 1 or higher and a minimum variance of 60 percent signify unidimensionality or communality of the scale. These findings confirm the unidimensionality of the scale used in the study and therefore the scale measures the traits of the constructs (Brown, 2001).