Once the research hypotheses were tested using hierarchical regression, further tests were carried out using PLS analysis to validate the research model and to examine the effects of the indicators in organization contexts and knowledge sharing on innovation and KMS. Given that the KM literature indicated that KMS and innovation can be crucial means for organizations to achieve performance and competitive advantage and are as such essential for organizations’
sustainability. Therefore for PLS analysis purposes, the researcher identified KMS and innovation as sustainability indices (see figure 4.1).
Measurement Model
Table 4.20 and Table 4.21 provided the tests’ results for the measurement model. The data showed that the measures of the constructs analyzed in this research were satisfactory in terms of their internal consistency reliability as indexed by composite reliability. Given the research’s constructs included reflective indicators, composite reliability was considered as a fundamental mean for assessing reliability. In general, guidelines provided by scholars (e.g. Lawson et al., 2009; Nunally & Bernstein, 1994) suggested constructs’ reliability should be equivalent to or greater than .70. In line with recommendations from scholars, this study’s results showed that composite reliabilities values exceed the threshold value (see Table 4.20).
Constructs’ Cronbach’s alpha coefficients were also examined to assess their reliability. The adequacy of the measurements was determined: 1) through reliability guidelines provided by scholars and 2) through comparison with composite reliability values. Although the results showed that some of the constructs did not fully meet the criteria for reliability established by Nunally (1994), the overall measurements were deemed as acceptable on the basis of the guidelines provided by Peterson (1994). This author indicated that cronbach’s alpha values of .60 represent the criterion in use. Cronbach’s alpha coefficients for knowledge sharing and sustainability indices clearly exceeded this threshold value (See Table 4.20).
In general composite reliability is considered a better estimate for reliability. While Cronbach’s alpha implicitly presumes that each item has the same weight, composite reliability depends on the actual loadings to determine the factor score. With that said, Cronbach’s alpha is
105
expected to provide lower bound estimates of reliability than composite reliability. On the basis of Fornell and Lacker (1981), the internal consistency for the measurement model was considered satisfactory given that the composite reliability was superior to the Cronbach alpha (see Table 4.20). For these reasons, the reliability of the measurement model was upheld.
The results also demonstrated satisfactory validity of the measures (see Table 4.21). In relations to convergent validity, the results were consistent with the recommendations of (Al-Busaidi et al., 2010; Li et al., 2009). Average variance extracted for all constructs exceeded the recommended threshold value of .50. In addition, the factor loadings for almost all constructs were found to exceed the threshold value suggested in (Hair et al., 2006; Li et al., 2009).
Although the factor loading for organization structure did not fulfill the threshold requirement for (Hair et al., 2006), the construct was considered as valid since its loading value was closed to .60. Average variance extracted values were used to evaluate the discriminant validity.
Discriminant validity of the measurement model was assessed by checking whether the correlations among the research variables were lower than the square root of the average variance extracted ( AVE ). The results in Table 4.22 showed that the square root of each AVE was greater than the off-diagonal correlations. This suggested acceptable discriminant validity among research variables.
Table 4.20
Measurement Model Reliability Analysis
Constructs Communality Composite
reliability (CR)
Cronbach’s Alpha
R2 Redundancy
Organization contexts .553 .828 .725
Knowledge sharing .754 .859 .674 .215 .161
Sustainability indices .744 .853 .656 .555 .380
Note: Sustainability indices include KMS and innovation.
106 Table 4.21
Measurement Model Convergent Validity Analysis
Loadings
Organization contexts Information technology (.797), top management support (.822), collaborative culture (.782), organization structure (.537).
AVE=.553
Knowledge Sharing Explicit knowledge sharing (.854), Tacit knowledge sharing (.882)
AVE=.754
Sustainability Indices KMS (.877), Innovation (.847) AVE=.744
Note: Sustainability indices include KMS and innovation.
Table 4.22
Correlation Matrix and Measurement Model Discriminant Validity Analysis
Constructs Mean SD 1 2 3
1 Organization contexts 3.405 .540 .744
2 Knowledge sharing 3.418 .541 .528** .868
3 Sustainability indices 3.276 .502 .718** .558** .863
Note: Sustainability indices include KMS and innovation. Final results for discriminant validity tests are reported on the diagonal line and in bold.
107 Structural Model
Having satisfied the requirement arising from measurement issues, the structural model in Figure 4.2 was subsequently tested. Figure 4.2 shows the variance explained (R2) in the dependent constructs and the path coefficients (β) and their corresponding t-values and the goodness of fit (GoF) for the model.
The results of the multivariate test of the structural model are presented in Table 4.20 and 4.22. Table 4.20 showed that the structural model as whole explained 21.5 and 55.5% of the variance in knowledge sharing and sustainability indices (i.e. KMS and innovation), respectively.
The results also showed that all independent variables (i.e. information technology, top management support, collaborative culture, and organization structure) including the mediator variable (i.e. knowledge sharing types) explained 55.5% of the variance in the impact of innovation and KMS success. On the basis of the guidelines provided in Karim (2009) for R2 relevance, it can be concluded that this model explained a substantial percentage of the variance in organization contexts and knowledge sharing impact on KMS and innovation (R2=55.5). This fulfilled the basic premise for the study which implied that organization contexts and knowledge sharing are important for innovation and KMS success.
Consistent with this research’s expectations, organizations contexts had a positive and significant effect on KMS and innovation. Although organization contexts was also found to have a positive effect on knowledge sharing, and that knowledge sharing had a positive effect on KMS and innovation, these results were inconsistent with this study’s expectations due to the fact the effects of organization contexts on knowledge sharing, and those of knowledge sharing on KMS and innovation were lower than expected. For example, organization contexts only explained 21.5 % percent of the variance in knowledge sharing. Although this R2 value is acceptable on the basis of Falk and Miller (2002) guidelines which suggested that an explanatory power of R2 value greater than 10 percent is adequate, however, it can be considered somewhat moderate based on the guidelines provided in Karim (2009) which indicated that the exploratory power of an R2 value of .33 and .19 should be considered as moderate and weak respectively.
Referring to the effects of organization contexts on knowledge sharing and KMS and innovation and the effects of knowledge sharing on KMS and innovation, the path coefficients
108
and their respective t-values are examined. With regards to path coefficients, Bontis and Fitz-enz (2002) suggested that path coefficients values are more accurate than correlations since they account for mediating and indirect causal paths. These authors also noted that each path coefficient value should fall between the range of -1.00 and +1.00. In addition, as shown in Table 4.23 and Figure 4.2, the path coefficients for this study’s variables were greater than .20.
Consequently they were considered as meaningful on the basis of guidelines provided by Chin (1998).
These paths also gave an overview of the magnitude of direct effects, indirect effects (i.e., the effect of the independent variables on the dependent variables through the mediators), and the effect coefficients (i.e., total effect equal to the direct plus indirect effects). Through the analysis of these direct and indirect relationships, the link between the research variables can be further assessed and understood. As showed in Figure 4.1, the paths that resulted from the regression equations provided in Table 4.23 indicated the association between organization contexts and KMS and innovation can be analyzed: 1) in terms of directness; 2) in relation to the form of effects and 3) in terms of the extent of effect. From table 4.24, it should also be noted that knowledge sharing had direct effect on KMS and innovation. In contrast, organization contexts besides of having a direct effect on KMS and innovation had also an indirect effect on these dependent variables through knowledge sharing.
With regards to the form of effect, the total effect can be discussed as follow. Since knowledge sharing included both tacit and explicit knowledge sharing, the indirect effect of organization contexts on KMS and innovation through explicit knowledge sharing was .082 leading to the total effect of .708 (direct effect + indirect effect= .626+.082). The indirect effect of organization contexts on KMS and innovation through tacit knowledge sharing was .084 leading to the total effect of .710 (direct effect + indirect effect=.626+.084). In relations to the extent of organization contexts and knowledge sharing effects on KMS and innovation, the results showed that organization contexts and knowledge sharing direct and positive associations with the sustainability indices were more significant. Although weaker, their indirect associations with KMS and innovation were significant.
109
From Table 4.24, the results related to the total effect showed that the indirect effects of organization contexts on KMS and innovation neither negated nor diminished their associated direct effects. In conclusion, organization contexts had an impact on the sustainability indices and the fact that these independent variables had also indirect links with KMS and innovation confirmed this study’s assumptions and provided adequate evidence that the relationships between organization contexts and sustainability indices (i.e., KMS and innovation) were mediated by knowledge sharing.
In order to further test the structural model and to examine the research hypotheses, t- values were generated through bootstrapping technique. Bootstrapping technique was utilized with a re-sampling of 900 (generated from the original data set) to test the significance of the PLS estimates of path coefficients. Table 4.23 showed that all path coefficients were positive and significant at (p<.001) and that all hypotheses were supported. From this table, it can be seen that the formulated hypotheses were accepted:
Hypothesis 1: Organization contexts have a positive effect on KMS.
Hypothesis 2: Organization contexts have a positive effect on innovation.
Hypothesis3: Organization contexts have a positive effect on knowledge sharing.
Hypothesis 4: Knowledge sharing has a positive effect on KMS.
Hypothesis 5: Knowledge sharing has a positive effect on innovation.
Given that the path from organization contexts to knowledge sharing (3.761, p<.001) was significant and the path from organization contexts to innovation and KMS (6.767, p<.001) including the path from knowledge sharing to innovation and KMS (2.256, p<.001) were also significant, it was concluded that knowledge sharing had a mediating effect between the relationship of organization contexts and innovation. It was also concluded that knowledge sharing had a mediating effect between the relationship of organization contexts and KMS.
Therefore Hypotheses 6 and 7 were also supported.
Overall the results showed that organization contexts (i.e. information technology, top management support, collaborative culture, and organization structure) had a positive effect on
110
knowledge sharing (i.e. tacit and explicit). In addition, it also showed that both types of knowledge sharing have a positive effect on KMS and innovation and organization contexts together with knowledge sharing types had a positive effect on innovation and KMS. In line with Figure 4.2 and Table 4.23, two structural equations can be reported.
Knowledge sharing (KS) =.463*OC +ε.
Since sustainability indices encompass KMS and innovation, the corresponding structural equation that showed the effect of organization contexts and knowledge sharing on KMS and innovation can be written as:
1) Sustainability indices (SI) =. 626*OC + .208*KS + ε.
At last, when analyzing the quality of the model, three indicators were taken into account 1) communality index and average communality; 2) average redundancy; 3) goodness of fit (GoF).
As shown in Table 4.20, results for communality indexes indicated that the quality of the measurement model for each block of indicators were adequate. In addition, the average communality value was equal to .684. This showed that all constructs, on average, reflected their nature through their indicators at a good level. Values for both communality indexes and average communality exceeded .50. Moreover, scores for communality indexes were similar to those of average variance extracted as discussed in (Pauwels et al., 2009).
The average redundancies at the level of .161 and .380 suggested that the constructs in each path of the structural model successfully reflected their role toward the outcome indicators at an acceptable level (see Table 4.20). Finally, the goodness of fit of the model was determined by applying the Tenenhaus et al. (2005)’s goodness of fit measure. The square root of the product between average communality and average R2 showed that the GoF value for this model was equal 0.684×0.385 = .513. This value clearly fell between the range of 0 and 1 as recommended in Karim (2009). In addition, it also indicated that the structural model could satisfactory predict the sustainability indices.
111 Table 4.23
PLS Hypotheses Testing Results
Path β-
path
Adj. t-value
Sig .
Direction Support for hypotheses
GoF
Organization Contexts →Knowledge Sharing .463 3.761 *** + Yes
Organization Contexts →KMS, Innovation .626 6.767 *** + Yes .513 Knowledge Sharing →KMS, Innovation .208 2.256 *** + Yes
Note: *** p <.001.
Table 4.24
Effects of Organization Contexts on KMS and Innovation
Path Direct
Effect
Indirect Effect
Total Effect
Organization Contexts →Knowledge Sharing Organization Contexts→ KMS, Innovation
.463 .626
.463 .626 Organization Contexts →EKS→KMS, Innovation .463x.854x.208=.082 .708 Organization Contexts →TKS→KMS, Innovation .463x.882x.208=.084 .710
Knowledge Sharing →KMS and Innovation .208 .208
112 Organization Contexts
Knowledge Sharing R2=.215
Sustainability Indices R2=.555 Information Technology
.797*** (14.261)
Top Management Support .822*** (8.227)
Collaborative Culture .782*** (7.274)
Organization Structure .537*** (2.798)
Explicit Knowledge Sharing .854*** (9.529) Tacit Knowledge Sharing
.882*** (11.134)
KMS .847*** (25.508)
Innovation .877*** (15.553) .463 *** (3.761)
.208*** (2.256)
.626*** (6.767)
GoF= .513 CR= .828
AVE= .553
CR= .859 AVE= .754
CR= .853 AVE= .744
Figure 4.1 Structural model’s empirical results
113