KNOWLEDGE


Figure 6 shows that the CFA model focuses solely on the interrelation between AKB factors and their measured variables. The model determines the goodness-of-fit between the factors in the hypothesised model and the sample data. The factor loading between each latent variable and its observed variables is important to be higher, i.e., ATTITUDE and its observed variables ATT3, ATT4, ATT5 and ATT6, with 0.03, 0.26, 0.12 and 0.22, respectively, are very low, which might be problematic, as they should be closer to 1 in order to achieve goodness of fit.

$$\text{t} < 12 \times \text{s} (\text{s} + 1)$$

The above equation, if confirmed, indicates that the model in Figure 6 is over identified:

t = items to be identified= 70 (35 'e' + 32 factor loading + 3 latent variables)s = number of observed variables (35)

> 70 < 630

Based on the above result, it is confirmed that the model (Figure 6) is over identified, meaning that we have more than enough observed variables (35) to identify unobserved items (70), and therefore we have the possibility of eliminating some if we need to, in order to achieve the best model fit.

#### *4.2. Conceptual Measurement Model Evaluation*

The list given below explains the acceptable and good fit data ranges [58,61] that need to be achieved in order to confirm that a model is a good fit:


The significance of the interrelationships among variables in the measurement/hypothesised model was tested in AMOS software in order to review the reliability and commonality of such a model. The results are presented in Table 2.


**Table 2.** Conceptual measurement model fit.

The list below checks whether the parameters in Table 2 related to the data for Figure 6 are within range and are a good fit, as explained above:


All of the figures in Table 2 indicate that Figure 6 is not a well-fitting measurement model, meaning that none of the observed variables/indicators have strong interrelationships with their own latent variables AKB.

#### *4.3. Conceptual Measurement Model Modification*

Based on the results given above, the measurement model was not a good fit, and needed to be modified. To achieve the best fitting measurement model, three main steps [62] were carried out, as follows:



**Table 3.** Modification Indices ≥11 for Step 2 modifications of observed variables interrelationships

The path coefficient and GOF sometimes reveal the need to modify models in SEM, which can result in the selection of the best fitting model falling within theoretical expectation and satisfying the GOF measures [42].
