Ordering of components of Green Supply Chain Practices jointly impacting the individual components of Green Supply Chain Performance – An Empirical Study of the Indian Automobile Manufacturing Sector

This paper establishes the order in which five identified green supply chain practices jointly impact ten identified individual component measures of green supply chain performance with reference to the automobile manufacturing sector of India. This research paper is an extension of the research work done by [1]. The purpose of this research paper is to test the hypotheses developed by [1]. Further the joint impact of Green Supply Chain Practices on individual components of Green Supply Chain Performance has been established by means of ten multiple regression models. Consistently the ten multiple regression models that were developed established that there is a definite ordering of the five Green Supply Chain Practices while jointly impacting each of the ten component measures of Green Supply Chain Performance individually. These findings would enable practicing managers in the automobile manufacturing sector of India to take decisions related to implementation of green supply chain practices which would result in enhancing a particular green supply chain performance measure. This information regarding implementation of Green Supply Chain Practices would be very handy as it has financial and policy making implications.


INTRODUCTION
The research problem here is to test sixty-one hypotheses out of which fifty have been developed by [1]  Literature that studies the impact of Green Supply Chain Practices on Green Supply Chain Performance measures is on the rise. Few of the studies that have addressed this linkage are as follows: [1]; [2]; [3]; [4]; [5]; [6]; [7].
It has been established by the existing literature that Green Supply Chain Practices have an impact on measures of GSC performance [8] only at a broad level. Not many studies have focused identifying the exact joint impact of Green Supply Chain Practices on particular component measures of Green Supply Chain Performance [4], [5], [6], [7]. Further the existing The data that was collected on administering the questionnaire on the respondents was entered in an EXCEL sheet manually by coding the responses on a 5-point balanced Likert scale as 1, 2 3 4 and 5. The data was subsequently transferred to statistical analysis software SASS for the analysis. The descriptive statistics of the data collected is shown in the Table 2. In order to evaluate the reliability of the data collected, Cronbach Coefficient Alpha was evaluated for each of the sub-constructs in the study. Table 3 shows the various sub-constructs involved in the study along with the corresponding value of the Cronbach Coefficient Alpha. This coefficient is a measure of reliability. Normally values starting from around 0.7 and going upwards are considered to indicate a good reliability. By reviewing the Cronbach Coefficient Alpha for the sub-constructs shown in the table 3, it is observed that the questionnaire is reliable to scale all the fifteen items. Sample size was 50 respondents for this pilot study whereas for getting a true indication of reliability, a sample size of around 100 respondents is needed. Accordingly, 103 samples were taken for this study during the major survey so that better conclusion could be drawn. The Cronbach Coefficient Alpha for Green Supply Chain Execution (Supply Loops) is less but it has got strong support of existing literature in its favour; so it has been retained [9].

FACTOR ANALYSIS OF THE DATA COLLECTED DURING THE PILOT STUDY
Confirmatory Factor Analysis was conducted on the variables constituting the sub-constructs DFE [10], EC [11], LCA [12], PP [13] and RL [14]. Confirmatory Factor Analysis was also conducted on the variables constituting the sub-constructs GSCPLAN [15], GSCPROC [16], GSCEXPROD [17], GSCEXLOG [18], GSCEXPACK [19], GSCEXMARK [20], GSCEXSL [21], CM [22], GSCMIG [23] and GSCCI [24] in a similar manner. This helped in identifying the factors and also in establishing the communality estimates for or each of the sub-constructs in the questionnaire. By sorting the component variables of each sub-construct in descending order of value of their communality estimates, it was possible to establish the order of contribution of component variables constituting each sub-construct [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23] and [24]. The major survey consisted of 103 samples including the 50 samples taken during the pilot survey. The ten sub-constructs of the construct GSC Performance (Green Supply Chain Performance) and the five sub-constructs of the construct GSC Practices (Green Supply Chain Practices) used in correlation analysis are shown in Table 4 in their abbreviated form. The descriptive statistics of the data collected and scaled during the major survey is shown in the Table 5.  Table 6 shows the Pearson's correlation coefficient between each of the five components of green supply chain practices and each of the ten components of green supply chain performance. Accordingly, in all fifty associations were identified for a co-relational study. On the basis of Table 6 it is possible to test the association between the fifty pairs of subconstructs and hence test the fifty hypotheses which have been framed [1]. Also additionally eleven hypotheses pertaining to the ordering and the joint influence of components of GSC Practices on individual component measures of GSC Performance have been framed. Table 7 shows all the sixty-one hypotheses to be tested in their null and alternate form. Also it shows the decision of accepting or rejecting the hypotheses on the basis of correlation coefficient and/or regression analysis. For values of p less than 0.05 the null hypotheses have been rejected else they have been accepted. The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 8, Table 9, The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 11, Table 12,  Table 13, Table 39, Table 41 and Figure 13. The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 14, Table 15,  Table 16, Table 39, Table 41 Reject null hypothesis The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 17, Table 18, Table 19, Table 39, Table 41 and Figure 15.

H34a
There is a definite order in which the five Green Supply  The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 20, Table 21,  Table 22, Table 39, Table 41 and Figure 16. The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 23, Table 24,  Table 25, Table 39, Table 41 and Figure 17.

H36a
There is a definite order in which the five Green  The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 26, Table 27,  Table 28, Table 39, Table 41 and Figure 18. The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 6, Table 32, Table 33,  Table 34, Table 39, Table 41 and Figure 20. The null hypothesis is rejected and the alternate hypothesis is accepted on the basis of the evidence provided by Table 10, Table 13, Table  16, Table 19, Table 22, Table  25, Table 28, Table 31, Table  34, Table 37, Figure 12 and Figure 21.

H41a
There is a definite order in which components of Green Supply Chain Practices jointly impact the components of Green Supply Chain Performance.
Accept alternate hypothesis From the first fifty hypotheses i.e. hypotheses from serial number 1 to serial number 50 of Table 7 it is evident that on the basis of correlation analysis the five components of green supply chain practices are individually associated in varying degrees with the ten subconstruct components of green supply chain performance (Hypotheses H1 to H30). Also it is evident from table 7 that the five green supply chain practices jointly impact individual component measures of Green Supply Chain Performance (Hypotheses H31 to H40). Finally hypothesis H41 in table 7 makes it evident that there is a definite order of influence of the five green supply chain practices on individual component measures of Green Supply Chain Performance.

REGRESSION ANALYSIS
Ten dependent sub-constructs and five independent sub-constructs were identified. Accordingly ten models and fifty (10 x 5 = 50) hypotheses emerged for doing regression analysis. Each of these models was tested one by one for studying the joint impact of the independent sub-constructs (i.e. components of GSC Practices) on a particular dependent subconstruct (component measures of GSC Performance). Accordingly, all the fifty hypotheses were stated in their null and alternate form as shown in table 7 for testing them by using multiple regression analysis. Also additional eleven hypotheses pertaining to the ordering of components of GSC Practices jointly influencing the component measures of GSC Performance are stated in their null and alternate form in table 7. In all sixty-one hypotheses are put to test.

Model 1. Green Supply Chain Planning (GSCPLAN)
Model 1 is associated with the five hypotheses namely H1, H7, H13, H19 and H25 wherein the dependent construct is GSCPLAN (Green Supply Chain Planning) and the independent subconstructs are EC, PP, RL, LCA and DFE. Model 1 is depicted in figure 2.

Figure 2. Model 1-Impact of GSC Practices on GSCPLAN
The summary of the multiple regression output for model 1 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCPLAN, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 1, the R 2 = 0.9606 which means that 96.06 % of the variance of the dependent construct GSCPLAN is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 8 shows the analysis of variance for model 1. Table 9 shows the computation of R 2 value for model 1. Table 10 shows the computation of parameter estimates for model 1. Thus, the hypotheses H1a, H7a, H13a, H19a and H25a are substantiated. Since some of the parameter estimates are negative, as shown in the table 10, there appears to be an existence of multicollinearity. The effect of multicollinearity can be removed by using Principal Component Regression. On applying Principal Component Regression the centered and scaled data is as shown in the figure

Model 2. Green Supply Chain Procurement (GSCPROC)
Model 2 is associated with the five hypotheses namely H2, H8, H14, H20 and H26 wherein the dependent construct is GSCPROC (Green Supply Chain Procurement) and the independent subconstructs are EC, PP, RL, LCA and DFE. Model 2 is depicted in figure 3. The summary of the multiple regression output for model 2 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCPROC, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 2, the R 2 = 0.9901 which means that 99.01 % of the variance of the dependent construct GSCPROC is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 11 shows the analysis of variance for model 2. Table 12 shows the computation of R 2 value for model 2. Table 13

Model 3. Green Supply Chain Execution -Production (GSCEXPROD)
Model 3 is associated with the five hypotheses namely H3A, H9A, H15A, H21A and H27A wherein the dependent construct is GSCEXPROD (Green Supply Chain Execution-Production) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 3 is depicted in figure 4. The summary of the multiple regression output for model 3 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCEXPROD, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 3, the R 2 = 0.9561 which means that 95.61 % of the variance of the dependent construct GSCEXPROD is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 14 shows the analysis of variance for model 3. Table 15 shows the computation of R 2 value for model 3. Table 16

Model 4. Green Supply Chain Execution -Logistics (GSCEXLOG)
Model 4 is associated with the five hypotheses namely H3B, H9B, H15B, H21B and H27B wherein the dependent construct is GSCEXLOG (Green Supply Chain Execution-Logistics) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 4 is depicted in figure 5.

Figure 5. Model 4-Impact of GSC Practices on GSCEXLOG
The summary of the multiple regression output for model 4 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCEXLOG, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 4, the R 2 = 0.9050 which means that 90.50 % of the variance of the dependent construct GSCEXLOG is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 17 shows the analysis of variance for model 4. Table 18 shows the computation of R 2 value for model 4.

Model 5. Green Supply Chain Execution -Packaging (GSCEXPACK)
Model 5 is associated with the five hypotheses namely H3C, H9C, H15C, H21C and H27C wherein the dependent construct is GSCEXPACK (Green Supply Chain Execution-Packaging) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 5 is depicted in figure 6. The summary of the multiple regression output for model 5 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCEXPACK, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 5, the R 2 = 0.9286 which means that 92.86 % of the variance of the dependent construct GSCEXPACK is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 20 shows the analysis of variance for model 5. Table 21 shows the computation of R 2 value for model 5. Table 22

Model 6. Green Supply Chain Execution -Marketing (GSCEXMARK)
Model 6 is associated with the five hypotheses namely H3D, H9D, H15D, H21D and H27D wherein the dependent construct is GSCEXMARK (Green Supply Chain Execution-Marketing) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 6 is depicted in figure 7.

Figure 7. Model 6-Impact of GSC Practices on GSCEXMARK
The summary of the multiple regression output for model 6 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCEXMARK, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 6, the R 2 = 0.9403 which means that 94.03 % of the variance of the dependent construct GSCEXMARK is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 23 shows the analysis of variance for model 6. Table 24 shows the computation of R 2 value for model 6. Table 25

Model 7. Green Supply Chain Execution -Supply Loops (GSCEXSL)
Model 7 is associated with the five hypotheses namely H3E, H9E, H15E, H21E and H27E wherein the dependent construct is GSCEXSL (Green Supply Chain Execution-Supply Loops) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 7 is depicted in figure 8.

Figure 8. Model 7-Impact of GSC Practices on GSCEXSL
The summary of the multiple regression output for model 7 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCEXSL, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 7, the R 2 = 0.8792 which means that 87.92 % of the variance of the dependent construct GSCEXSL is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 26 shows the analysis of variance for model 7. Table 27 shows the computation of R 2 value for model 7. Table 28

Model 8. Carbon Management (CM)
Model 8 is associated with the five hypotheses namely H4, H10, H16, H22 and H28 wherein the dependent construct is CM (Carbon management) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 8 is depicted in figure 9. The summary of the multiple regression output for model 8 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct CM, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 8, the R 2 = 0.9005 which means that 90.05 % of the variance of the dependent construct CM is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table  29 shows the analysis of variance for model 8. Table 30 shows the computation of R 2 value for model 8. Table 31

Model 9. Green Supply Chain Migration (GSCMIG)
Model 9 is associated with the five hypotheses namely H5, H11, H17, H23 and H29 wherein the dependent construct is GSCMIG (Green Supply Chain Migration) and the independent subconstructs are EC, PP, RL, LCA and DFE. Model 9 is depicted in figure 10. The summary of the multiple regression output for model 9 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCMIG, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 9, the R 2 = 0.8908 which means that 89.08 % of the variance of the dependent construct GSCMIG is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table 32 shows the analysis of variance for model 9. Table 33 shows the computation of R 2 value for model 9. Table 34

Model 10. Green Supply Chain Continuous Improvement (GSCCI)
Model 10 is associated with the five hypotheses namely H6, H12, H18, H24 and H30 wherein the dependent construct is GSCCI (Green Supply Chain Continuous Improvement) and the independent sub-constructs are EC, PP, RL, LCA and DFE. Model 10 is depicted in figure 11. The summary of the multiple regression output for model 10 is as follows: When the five independent or predictor constructs namely EC, PP, RL, LCA and DFE are jointly regressed against the dependent or criterion construct GSCCI, which is interval scaled, the five individual correlations collapse into what is called as a multiple r or multiple correlation. The square of the multiple r which is also commonly known as R-square or R 2 is indicative of the amount of variance in the dependent construct explained jointly by the predictors. In the case of model 10, the R 2 = 0.9290 which means that 92.90 % of the variance of the dependent construct is significantly explained jointly by the predictors EC, PP, RL, LCA and DFE at a significance level of α = 0.05 (p < 0.0001), i.e., this does not hold true 0.0001 % of times. Table  35 shows the analysis of variance for model 10. Table 36 shows the computation of R 2 value for model 10.   Table 38 shows the details of Principal Component Regression which was performed. There were ten response variables and five predictor variables. Missing value was not needed and one factor was extracted. A total of 103 responses were obtained during the survey for the analysis.  Table 39 shows that the principal components collectively account for 72.5010 % of variation of the five dependent variables.  Table 40 shows the parameter estimates for the centered and scaled data pertaining to the GSC practices and GSC performance. Then the new (revised) parameter estimates are calculated.  The new (revised) parameter estimates are shown in the table 41 along with the intercept values. These new parameter estimates reveal that there is no effect of multicollinearity existing now. Hence these coefficients are dependable for building the ten regression equations. Accordingly, the regression equations for predicting the ten component measures of GSC performance using the five component measures of GSC practices are as follows:          Also the order of influence of each of the five components of GSC Practices on each of the ten components measures of GSC Performance is consistently in the descending order of influence of the GSC Practices namely LCA, EC, PP, DFE and RL. Also on the basis of communality estimates (h 2 ) of the components of each of the five GSC Practices and on the basis of the communality estimates of the components of each of the ten GSC Performance measures it is possible to rank the order of the variables constituting them as established in [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], The existing body of knowledge has established at a broad level that GSC Practices have an impact on GSC Performance. However, it has not been able to establish very conclusively as to which of these GSC Practices specifically impacts which of the GSC Performance measures. This study was set out to explore the unexplored linkages between GSC practices and component measures of GSC Performance. Several definitions of GSC Practices and GSC Performance emerged during a detailed literature review. But the impact of GSC Practices on GSC Performance measures using the combination of definitions of GSC practices and GSC Performance as used in this research study has not been explored before. The research study was set out to study the joint impact of five identified Green Supply Chain Practices on ten identified individual component measures of GSC Performance with reference to the Indian automobile manufacturing sector by means of an empirical study by administering a questionnaire on representatives of automobile manufacturing firms and plants.

PRINCIPAL COMPONENT REGRESSION
The study could establish the fact that each of the five GSC Practices has a significant positive correlation with each of the ten individual component measures of GSC Performance which is evident from the correlation coefficients computed between each of them. This finding is in line with the findings in existing literature also. Also from this it was possible to find the order in which the five GSC Practices are correlated with each of the ten individual component measures of GSC Performance. Looking at it in the other way the study could also establish the order in which each of the ten GSC Performance measures is correlated with each of the five GSC Practices. This is a unique finding made by this study.
On doing regression analysis it was possible to establish ten regression equations each used to establish the joint impact of each of the five GSC Practices on each of the individual component measures of GSC Performance. In short since there are ten measures of GSC Performance, ten multiple linear regression equations were obtained. Each regression equation helps to establish a particular GSC Performance measure. These ten linear multiple regression equations can be used to predict the individual GSC Performance measures.
A closer look at the coefficients of these ten linear multiple regression equations revealed these coefficients or the parameter estimates follow a particular pattern. In each of the ten multiple regression equations it was observed that the parameter estimates had a particular hierarchy. The parameter estimates were consistently highest for LCA followed by EC followed by PP followed by DFE followed by RL. This means that whenever the five GSC Practices jointly impact GSC Performance measures, they do so in a particular order. And this order is consistent when applied to each of the ten GSC Performance measures. So it can be conclusively stated that there is a definite order in which each of the five GSC Practices will jointly impact each of the ten individual component measures of GSC Performance. This is a finding which has not been established by existing research. This is one of the key findings of this research work.
Accordingly, in line with the above discussion, in all ten models or multiple linear regression equations were obtained. The goodness of a model is measured by its R 2 value. The R 2 value is a measure of the amount of variance in the dependent construct explained jointly by the predictors or independent constructs. In the case of this research the dependent constructs are the ten individual GSC Performance measures and the predictors or the independent constructs are the five GSC Practices. By ranking the R 2 values of the ten multiple regression equations it is possible to know which model (GSC Performance measure) is able to explain the joint variation of the five GSC Practices the most and also which model or (GSC Performance measure) is able to explain the joint variation of the five GSC Practices the least. Accordingly, we can get the ordering of GSC Performance measures as regards their ability to explain the joint variation of the five GSC Practices. This order is as follows: GSCPROC with a R 2 value of 0.9901; followed by GSCPLAN with a R 2 value of 0.9606; followed by GSCEXPROD with a R 2 value of 0.9561; followed by GSCEXMARK with a R 2 value of 0.9403; followed by GSCCI with a R 2 value of 0.9290; followed by GSCEXPACK with a R 2 value of 0.9286; followed by GSCEXLOG with a R 2 value of 0.9050; followed by CM with a R 2 value of 0.9005; followed by GSCMIG with a R 2 value of 0.8908; followed by GSCEXSL with a R 2 value of 0.8792. This finding is a bye product of this research, but it has interesting insights. Practitioners can make use of this ordering of GSC Performance measures based on R 2 to focus on improving a particular component measure of GSC Performance. It is important to know this ordering as it helps in prioritizing the GSC Performance improvement projects to be taken up first. Prioritizing is needed because most of the projects have financial implications associated with them.
The findings of this research also add to the existing body of knowledge as these are unique findings.
Based on the value of communality estimates (h2) of the variables constituting each construct it is possible to conclude about how much of each variable is accounted for by underlying factors taken together. Accordingly it is possible to arrive at the order of contribution of the variables constituting each of the sub-constructs of GSC Practices and GSC Performance.