4.2. Reflective Measurement Models
In the reflective measurement model evaluation (
Table 3), the factor loading, a value higher than 0.707, was examined to ensure the construct could explain over 50% of the indicator variance. The results met the criterion, providing acceptable reliability for each observable variable. After that, the internal consistency reliability was considered using Cronbach’s alpha (α), rho_A (ρA), and rho_c (ρc or Jöreskog’s composite reliability); rho_A between the values 0.700 and 0.900 presents the internal consistency reliability of a construct [
71]. The finding was within the given criterion; therefore, every indicator had correlations to determine the identical construct. In the following step, convergent validity was evaluated from the average variance extracted (AVE), which should be higher than 0.500. The findings followed the criteria; therefore, the construct could explain over 50% of the variance of its indicators. Finally, the discriminant validity assessment (
Table 4) ensured that the different constructs measured different characteristics. The AVE values in the Fornell–Larcker matrix presented in the diagonal were higher than the correlation values in the same column and row. Furthermore, every heterotrait–monotrait (HTMT) ratio of the correlations was less than 0.850; therefore, each construct in the model measured different characteristics.
For the predictive power of the structural model, three instruments were evaluated: R
2, Q
2, and PLSpredict (
Table 5). The R
2 value was considered from a guideline: R
2 values of 0.75, 0.50, and 0.25 can be assessed as substantial, moderate, and weak, respectively [
71]. The results of R
2 showed that the explanatory power of DT was extremely weak, while those of OSI and HPO were weak. However, the R
2 value only indicated the model’s in-sample explanatory power. Accordingly, it was impossible to estimate the predictive power using only R
2 [
72].
4.3. Structural Model
Structural model coefficients explaining the relationship of couple constructs originate from assessing regression equations; therefore, the collinearity had to be considered to ensure no bias. Predictor variables could not independently predict the dependent variable value when the regression models correlated. A variance inflation factor (VIF) less than 3 was utilized for the investigation. According to the results, every VIF value (
Table 3) met the defined criterion.
The predictive accuracy assessment of the model can be considered from the Q
2 value estimated using the blindfold technique. This method predicted the data points removed for all variables. The differentiation between the predicted and original values was cracked to a Q
2 value, indicating predictive accuracy. The Q
2 value combined the aspects of out-of-sample prediction and in-sample explanatory power [
73]. The Q
2 combined the in-sample explanatory power and out-of-sample prediction perspectives [
73]. The value was evaluated from a guideline: Q
2 values of 0.5, 0.25, and 0 are high, medium, and low predictive relevance of the PLS-path model, respectively [
71]. The findings of Q
2 showed that the predictive accuracy of OSI was medium, but those of DT and HPO were low.
PLSpredict was developed for out-of-sample-based predictions using PLS-SEM. Initially, the Q
2predict value, evaluated to confirm that the model outperformed the most naïve benchmark, must be greater than zero. Considering that the prediction errors were highly unsymmetrically distributed, the mean absolute error (MAE) was used to compare the values between PLS-SEM and the naïve benchmark (linear regression model, LM) [
74]. The results showed that the predictive power of DT was high, but those of OSI and HPO were medium.
4.4. Hypothesis Testing
This study used PLS-SEM to compute the structural equation to test the hypotheses on the relationships and influence of KBDCs, DT, OSI, and HPO. The results of the structural model are shown in
Figure 1. The findings (
Table 6) show that KBDCs had a significant positive effect on DT (β = 0.388, t = 4.913, and
p = 0.000). Thus, H1 was supported. The principal cause of DT’s failure was a lack of knowledge of the progress of digital strategies [
38]. The results empirically confirmed that, the higher the KBDCs, the better the DT, irrespective of creating or deploying digital technology in digital artifacts, platforms, or infrastructure. This concurs with Alvarenga, Matos, Godina, and Matias’s [
75] work. They studied the relationship between the implementation of DT and the use of KM practices in public organizations and concluded that KM was a critical factor in the success of DT. The bibliometric analysis by Di Vaio, Palladino, Pezzi, and Kalisz [
76] suggested that knowledge management systems (KMS) were crucial in ensuring the optimization of technologies and resources. Simultaneously, the technologies adopted in the KMS develop the processes to be optimized. Furthermore, the digital platform fosters open innovation activities to develop novel qualitative features and highly reduces the overall cost of innovation [
77].
KBDCs had a significant positive effect on OSI (β = 0.350, t = 5.285, and
p = 0.000). Thus, H2 was supported. The results supported the idea that organizational KBDCs assist in the development of OSI. Absorptive and generational capabilities promote the acquisition of new knowledge. Storage capability encourages the maintenance of complete knowledge and facilitates rapid access to knowledge. These capabilities reinforce learning for generating OSI. Adaptation capability implies organizing the application of knowledge, adjusting strategic directions, and coping with new challenges. This concept is reinforced by prior studies concluding that KMS plays a vital role in supporting managers in decision-making processes [
76]. According to an empirical study, dynamic KM capability significantly affected SI [
8]. In addition, KBDCs had an indirect significant positive effect on OSI (β = 0.188, t = 3.440, and
p = 0.000), and DT functioned as a partial mediator of this relationship; this follows the conclusions of Di Vaio et al. [
76] that digital technologies provide efficient KM; therefore, companies can enhance the use of KMS to support strategies and decision making.
DT had a significant positive effect on OSI (β = 0.484, t =6.885, and
p = 0.000). Thus, H3 was supported. DT facilitates the creation of OSI and decision making. According to a study on the impact of DT, digital marketing encouraged strategies for promotion, brand positioning, and e-business development [
78]. Shen, Hua, Huang, Ebstein, and Yu [
79] suggested a KM-based digital platform for strategic solutions in property management. Thus, existing literature can interpret this in two ways: (1) DT contributes to choosing a consistent strategy, or (2) DT influences OSI to create a strategy. Further studies are required to investigate this; many studies have confirmed the importance of the DT strategy. For example, a study on the effects of DT on the Indian manufacturing industry showed that strategic alignment positively affected DT and performance; this highlights an understanding of the organizational process for designing and translating an appropriate DT strategy prior to transformation [
80]. Although the current research on the relationship between DT and OSI is ambiguous, the results serve as a starting point for extending the literature in this area.
KBDCs had a significant positive effect on HPO (β = 0.176, t = 2.190, and
p = 0.014). Thus, H4 was supported. The findings correlated with empirical research in which Sayyadi [
81] concluded that managers could build an HPO through KM to develop a good understanding. The organization must transform isolated knowledge into a cohesive knowledge base to introduce innovations, support decision making, solve business problems, provide inputs for training, automate business routines, and improve organizational efficiency [
17]. In addition, KBDCs had an indirect significant positive effect on HPO (β = 0.237, t = 3.583, and
p = 0.000), and OSI functioned as the partial mediator of this relationship. It can be interpreted that KBDCs directly support HPOs and produce knowledge that stimulates the creation of strategies that result in the HPO. Although no study confirms the effect of this indirect correlation, the results provide new knowledge that can be expanded.
DT had a significant positive effect on HPO (β = 0.247, t = 2.597, and
p = 0.005). Thus, H5 was supported; this was an expected result, as the usage of innovation and technology was considered to describe HPO. Therefore, the implementation of DT is an accelerator for introducing them into the organization. According to Wang, Feng, Zhang, and Li [
82], DT significantly influences organizational functions such as IT, information systems, and business operations. These effects change in business models, structures, and processes. Thus, the implementation of DT is strategic to improve organizational performance. In addition, DT positively correlates with short- and long-term financial performance. The results correlated with Sousa-Zomer, Neely, and Martinez’s [
83] work in that DT capability directly affected firm performance. Conversely, the empirical results revealed that DT had no significant direct effect on firm performance; however, it had an indirect effect (smart technologies are mediators). In addition, the results showed that DT directly affected the development and usability of smart technologies applied to improve organizational performance [
84]. Furthermore, DT had a significant positive indirect effect on HPO (β = 0.127, t = 2.376, and
p = 0.009), and OSI functioned as the partial mediator of this relationship. The study observed a direct effect between DT, OSI, and HPO. The indirect effect can expand the learning that DT is an organizational transformation that leads to more diversified strategy approaches to develop into HPO.
OSI had a significant positive effect on HPO (β = 0.263, t = 2.545, and
p = 0.005). Thus, H6 was supported. Using the right strategy at the right time ensures good organizational performance. Intuitive decision making occasionally leads to great strategies to achieve good organizational responsiveness and performance [
69]. Consistent with an empirical study, the results showed that SI capability significantly and positively affects firm performance. In this study, the researchers focused SI on personal capability. They explained that entrepreneurs should focus on their individual development to make decisions based on a state of mind that consciously thinks of processes that help acquire knowledge [
13]. However, the authors took a wide perspective on SI in this study. Although SI is an individual capability, an organization comprising employees with high SI will produce high OSI. Consequently, the organization can achieve high performance. The findings show that KBDC and DT will support the development of OSI and drive an organization towards HPO.