C. CONTEXTO TEATRAL: ALGUNAS CLAVES PARA CLARAMONTE
2. Acercamiento al uso cualitativo de la versificación dramática en El valiente
2.4 Conclusiones sobre el uso de cualitativo de la versificación
With the above standardized measurement model for each single construct, this section test the overall CFA model that fits all the nine constructs for this study. CFA model was run in AMOS version 26with a plugin “Pattern Matrix Builder” for single CFA model. The plugin automates the tedious job of creating a CFA from a pattern matrix by simply paste a principle component or pattern matrix from SPSS into the plugin window of AMOS version 26 and it automatically generate the single CFA model. All that need to be done is to rename the latent factors appropriately.
The model comprises the six Perceived Relative Advantage (PR) constructs, seven Perceived Compatibility (PCT) constructs, seven Perceived less Complexity (PCL) constructs, seven ICT infrastructure support characteristics (INF)constructs, six ICT Skills support characteristics (SS) constructs, five Perceived Competitive Pressure (PCP) constructs, five Perceived Pressure from Customers (PPC)constructs, five construct of the moderating variable ICT usage (ICTU) and six construct of the performance indicators (PI). CFA was run with maximum likelihood estimate in Amos software to determine its fitness. After the initial run, all the produced standardized regression weights were above 0.5 as a rule of thumb proposed by Hair et al. (2010) and significant at p ≤ 0.05 within the acceptable critical ratio (CR) of above ±1.96 and standardized residual covariance within the acceptable range (below 2.58).
In addition, the standardized model produced the following fit indices; CMIN/DF = 1.669, GFI = 0.753, TLI = 0.927, CFI = 0.931 and RMSEA= 0.053. This is not a good model fit based on the ration of GFI and RAMSEA which is below the recommended ration of GFI ≥ 0.90 by (Hoe, 2008). Hence the need for further refinement in order to achieve a model fit.
After observing the modification indices, the error variances suggested the need for removing item SS4 as its error terms appeared to be causing instability in the model. The following were the output after running the model for the second round; CMIN/DF = 1.629, GFI = 0.763, TLI = 0.930, CFI = 0.931 and RMSEA= 0.052.
These indicate slightly improvement however; poor fit is observed hence the need for further refinement in order to achieve a good model fit.
In order to improve the model, the researcher observed the modification indices which suggested the need to remove items PCT10 and PCT12 as they had high modification indices of 25,822 and 25,358 therefore causing instability of the model. The model was re-runed for the third time and it produced the following indices; CMIN/DF = 1.605, GFI = 0.773, TLI = 0.931, CFI = 0.936 and RMSEA= 0.051. This indicates poor model fit hence the need for further refinement in order to achieve a good model fit.
The researcher observed standardized residual covariance to ensure a good fit. Olivares and Shi (2017) recommended that, standardized residual covariance provide information about the source of misfit in poorly fitting models and they may be considered an alternative to modification indices with a value under 2.00 or 2.58. After checking the output of the standardized residual covariance data, PR2,INF4 and PCL6 had a greater value of 3 causing instability of the model. Therefore, they were deleted and the model was re-runed for the fourth time and it produced the following indices; CMIN/DF = 1.519, GFI = 0.796, TLI = 0.945, CFI = 0.950 and RMSEA= 0.047. This is a good fitting model based on Radzi and Hui et al.(2017) who recommended that, only four model fitting tests are required for acceptable results.
Likewise, Hoe (2008) recommends that a good model fit should achieve the following minimum fit indices; CFI (>0.90 indicates good fit), GFI ≥ 0.90, TLI >
0.90, RMSEA (< 0.08 indicates acceptable fit), and commonly used χ2 statistic (CMIN/DF ratio of 3 or less) in order to be considered fit.The value of GFI which is 0.796 is just below the recommended threshold of 0.90 but should be accepted because further improvement of the model could result in model over-fitting.Scholars provided some warning on the danger effect of over-fitting model based on trivial effects arising from capitalization on the chance factors. For examples, Ping, 2004 and MacDonald et al. (1988) suggested that, GFI has high chance of decreasing when the number of factors items included in a model increases. In view of this statistical evidence, it is concluded that the nine scales achieved adequate convergent validity, reliability, and unidimensionality (figure 4.14). Figure 4.12 presents the overall CFA model for this study and model estimates summarized in table 4.12
Figure 4. 12: Figure 4 12: Standardized Overall CFA Model
In addition to the fit indices of the CFA which showed good fitting model, the researcher observed the p-values which were all significant at p<0.05 as shown in Table 4.16.
Table 4.16: Standardized and Unstandardized Estimates for the Overall CFA Model
Path Unstandardized Estimate S.E. C.R. P Standardized Estimate
PCL1 <--- PCL 1.000 .957 PCL2 <--- PCL .996 .033 30.491 *** .932 PCL3 <--- PCL 1.027 .033 31.535 *** .939 PCL4 <--- PCL 1.013 .034 29.732 *** .927 PCL5 <--- PCL .961 .039 24.935 *** .886 PCL7 <--- PCL .968 .041 23.902 *** .875 PCT1 <--- PCT 1.000 .841 PCT2 <--- PCT 1.061 .055 19.393 *** .909 PCT5 <--- PCT 1.086 .047 23.037 *** .997 INF1 <--- INF 1.000 .907 INF2 <--- INF .973 .044 21.912 *** .885 INF3 <--- INF .980 .043 23.055 *** .902 INF6 <--- INF .916 .062 14.858 *** .735 INF7 <--- INF .707 .048 14.726 *** .731 INF8 <--- INF 1.009 .032 31.989 *** .998 PPC1 <--- PPC 1.000 .848 PPC2 <--- PPC 1.012 .056 18.151 *** .878 PPC4 <--- PPC .890 .073 12.260 *** .686 PPC7 <--- PPC .980 .068 14.315 *** .762 PPC8 <--- PPC .995 .045 21.880 *** .974 PI1 <--- PI 1.000 .723 PI3 <--- PI 1.027 .064 16.059 *** 1.005 PI4 <--- PI .957 .078 12.268 *** .772 PI5 <--- PI 1.066 .075 14.182 *** .885 PI6 <--- PI .986 .075 13.215 *** .828 SS1 <--- SS 1.000 .870 SS2 <--- SS .999 .061 16.365 *** .825 SS3 <--- SS 1.017 .058 17.539 *** .857 PR1 <--- PR 1.000 .577 PR5 <--- PR .870 .107 8.130 *** .643 PR6 <--- PR 1.311 .131 10.032 *** .974 PR7 <--- PR 1.134 .135 8.410 *** .675 PR8 <--- PR 1.251 .140 8.939 *** .740 PCP1 <--- PCP 1.000 .438 PCP3 <--- PCP 1.647 .225 7.304 *** .981 PCP5 <--- PCP 1.619 .230 7.034 *** .839 PCP6 <--- PCP 1.597 .230 6.932 *** .801 ICTU1 <--- ICTU 1.000 .692 ICTU2 <--- ICTU 1.165 .101 11.506 *** .814 ICTU4 <--- ICTU 1.208 .107 11.336 *** .801 ICTU5 <--- ICTU 1.290 .113 11.421 *** .808 ICTU6 <--- ICTU 1.286 .114 11.259 *** .795
Path Unstandardized Estimate S.E. C.R. P Standardized Estimate PCT7 <--- PCT .981 .075 13.106 *** .715 PCT8 <--- PCT 1.062 .063 16.961 *** .844 PI7 <--- PI .948 .081 11.701 *** .738 SS5 <--- SS 1.002 .058 17.180 *** .848 SS6 <--- SS 1.009 .060 16.703 *** .834 PCP7 <--- PCP 1.657 .239 6.922 *** .797