Professional Documents
Culture Documents
As part of measurement model evaluation, two items (OC4, OL2) were removed from the analysis
because of low factor loadings (<0.700) (Gefen and Straub, 2005). To test the reliability of the
constructs, the study used Cronbach’s alpha and composite reliability (CR). All the CRs were higher than
the recommended value of 0.700 (Wasco and Faraj,2005). Cronbach’s alpha of each construct exceeded
the 0.700 threshold. Convergent validity was acceptable because the average variance extracted (AVE)
was over 0.500. The results for reliability and validity along with the factor loadings for the items are
presented in (Table 1). Discriminant validity was assessed by Fornell-Larcker criterion, the table shows
the square root of AVE for the construct was greater the inter-construct correlation (see Table2).
Discriminant validity was also assessed by heterotrait-monotrait ratio of correlations (Henseler et
al.,2015), with values below the threshold of 0.900. Hence discriminant validity was established (see
Table 3).
IM OC OL OP
IM 0.801
OC 0.584 0.794
OL 0.656 0.617 0.832
OP 0.611 0.624 0.601 0.854
Note: Values in italic represent square root of AVE
IM OC OL OP
IM
OC 0.619
OL 0.691 0.674
OP 0.652 0.682 0.65
Structural Model
The structural model reflects the paths hypothesized in the research framework. A structural model is
assessed based on the R2, Q2 and significance of paths. The goodness of the model is determined by the
strength of each structural path determined by R2 value for the dependent variable (Briones Pefielver et
le.,2018) the value for R2 should be equal to or over 0.1 (Falk & Miller 1992). The result in Table 4 shows
that all values of R2 are over 0.1. Hence the predictive capability is established. Further Q2 establishes the
predictive relevance of the endogenous constructs. A Q2 greater than 0 shows that the model has
predictive relevance. The results show that there is significance in the prediction of the constructs (see
Table 4). Furthermore, the model fit was assessed using SRMR. The value of the SRMR was 0.070, this is
below the required value of .10, indicating acceptable model fit (Haier et al.,2016). Furthermore,
assessment of the goodness of fit, Hypothesis were tested to ascertain the significance of the
relationship. H1 evaluates whether IM has a significant impact on OP (β =.277, t=2.255, p=.025). Hence
H1 was supported. H2 evaluates whether IM has a significant impact on OC. (β =.584, t=8.007, p=.000).
Hence H2 was supported. H3 evaluates whether IM has a significant impact on OL. (β =.656, t=8.423,
p=.000). Hence H3 was supported. H4 evaluates whether OC has a significant impact on OP. (β =.330,
t=2.353, p=.019). Hence H4 was supported. H5 evaluates whether OL has a significant impact on OP. (β
=.216, t=1.947, p=.052). Hence H5 was not supported. Hypothesis testing results are summarized in
(Table 4)