46 Chapter IV Results and Analysis This chapter shows the results and discussion of the research findings. This chapter start with the results of hypothesis testing using structural equation model along with the analysis of the result regarding the effect of CSR attributes on Corporate Financial Performance (CFP) mediated by Innovation (I), Access to finance (AF) and customer satisfaction (CS), as well as to determine the effect of Economics Responsibility (ER), Legal Responsibility (LR), and Voluntary Responsibility (VR) to CSR. IV.1 Introduction This study aims to reveal the variables that influence Corporate Social Responsibility in SMEs and to determine the effect of Corporate Social Responsibility on financial performance with the variables that mediate it in SMEs in Indonesia. IV.2 Results IV.2.1 Pilot Study Pilot study is necessary to the development of a feasible full-scale study especially research that the questionnaire instrument have not test yet. In this study, the pilot study involved 55 respondents. The purpose of the pilot test in this study is to investigate the quality of questionnaire design, measurement scale, construct selection, and translation of indicator into measurement items are reliable and valid. Model measurement could be determined by examining construct and indicator (reliability) and convergent and discriminant (validity), the result of the pilot study presented in the Table IV.1. The construct reliability is evaluated based on the Cronbach alpha and composite reliability (CR) with 0.7 as its lower threshold. For indicator reliability, the minimum acceptable threshold of outer loading is 0.7. However, outer loadings that falls between 0.4 and 0.7 can be retained if the value of CR and average variance extracted (AVE) are already above the threshold. Besides that, the convergent validity assessed by the AVE values of each construct and they have to be above 0.5. 47 Table IV. 1 Pilot Study Results of the Outer Loadings, Cronbach Alpha, Composite Reliability, and AVE Indicator Loading Value Composite Reliability Average Variance Extracted (AVE) Cronbach's Alpha VR1 0.845 0.872 0.632 0.814 VR2 0.771 VR3 0.851 VR4 0.705 LR1 0.774 0.811 0,519 0.691 LR2 0.737 LR3 0.657 LR4 0.708 ER1 0.928 0.851 0.830 0.931 ER2 0.952 ER3 0.835 ER4 0.926 CSR1 0.739 0.879 0.725 0.869 CSR2 0.787 CSR3 0.935 CSR4 0.927 AF1 0.791 0.861 0.613 0.786 AF2 0.589 AF3 0.855 AF4 0.865 CS1 0.834 0.890 0.729 0.828 CS2 0.855 CS3 0.872 CS4 0.835 I1 0.757 0.894 0.628 0.853 I2 0.815 I3 0.821 I4 0.780 FP1 0.788 0.934 0.825 0.894 FP2 0.877 FP3 0.945 FP4 0.902 CFP1 0.712 0.740 0.585 0.841 CFP2 0.847 CFP3 0.857 CFP4 0.848 48 In this pilot test, it can be seen that there all of the criteria required for construct reliability, indicator reliability, and convergent validity were fulfilled. Even though there are some loading values that below 0.7. There are two loading value that around 0.5 and one around 0.6, but Cronbach alpha and construct reliability values that are larger than 0.7 with one Cronbach alpha that at the edge of 0.6 thresholds. In addition, the AVE values are all surpass 0.5, it meand that the chosen indicators and construct for this study might generate a consistent outcome. IV.2.2 Full-scale Sudy The result of hypothesis testing in this study is can be seen in the table IV.2. Table IV.2 present the result of outer loadings, internal consistency, convergent and discriminant reliability. It can be seen that there are two items with loading value around 0.5 and three items with loading value around 0.6 but the composite reliability for all associated construct was greater than 0.8 which indicates internal consistency and the model’s reliability. The convergent validity of the reflective measurement model was also tested, the AVE values was greater than 0.5 it considered to be acceptable. Table IV. 2 Outer Loading, Reliability, and Convergent Validity of Variables Items Outer Loading Cronbach alpha Compound Reliability AVE Volountary Responsibility 0.755 0.844 0.576 VR1 0.817 VR2 0.826 VR3 0.807 VR4 0.648 Legal responsibilities 0.701 0.769 0.625 LR 1 0.778 LR1 0.624 LR2 0.694 LR3 0.693 Economic responsibilities 0.730 0.804 0.512 ER1 0.637 ER2 0.787 ER3 0.855 49 ER4 0.759 Corporate social responsibilities 0.760 0.850 0.591 CSR1 0.602 CSR2 0.745 CSR3 0.881 CSR4 0.819 Access to financing 0.755 0.844 0.576 AF 1 0.759 AF 2 0.690 AF 3 0.797 AF 4 0.784 Customer satisfaction 0.719 0.829 0.557 CS 1 0.780 CS 2 0.858 CS 3 0.800 CS 4 0.784 Innovation 0.852 0.901 0.698 I 1 0.880 I 2 0.918 I 3 0.850 I 4 0.673 Company size, firm age, ownership 0.798 0.706 0.507 FP 1 0.759 FP 2 0.811 FP 3 0.757 FP 4 0.785 Corporate financial performance 0.677 0.808 0.528 CFP 1 0.850 CFP 2 0.774 CFP 3 0.805 CFP 4 0.839 In Table IV.3 reveals the discriminant validity value, it is the extent to which a construct is different from other constructs by empirical standards and in this study evaluated in accordance with Fornell-Larcker Criterion. (Fornell & Larcker, 1981). This technique compares the average root of the AVE variables with other latent variable correlations and the square root of each should be greater than its highest corresponding correlation coefficient. Table IV.3 highlights the AVE value square roots in bold and establishes discriminant validity is all met the requirement 50 Table IV. 3 Fornell-Larcker Criterion for Measurement Model Discriminant Validity Fornell-Larcker Criterion AF CFP CS CSR ER FP I LR VR AF 0.759 CFP 0.397 0.726 CS 0.610 0.486 0.746 CSR 0.563 0.354 0.609 0.769 ER 0.570 0.552 0.653 0.687 0.715 FP 0.214 0.524 0.202 0.157 0.206 0.638 I 0.685 0.490 0.593 0.766 0.580 0.269 0.836 LR 0.406 0.531 0.498 0.519 0.596 0.407 0.416 0.776 VR 0.376 0.282 0.466 0.493 0.451 0.145 0.398 0.479 0.778 Note: The bold values represent the square root of AVE, meanwhile the off diagonals show the correlation between construct The discriminant validity also can be assessed by seeing the items cross loading. Discriminant validity is achieved when an indicator’s loading on its assigned construct is higher than all of its cross-loadings with other construct (Hair et al., 2017). For example, from Table IV.4, item VR1 has the highest loading value with the appropriate construct "Volountary Responsibility" (0.817), while all VR1 cross-loading with other constructs is lower than 0.817. The results show that all items achieve discriminant validity because all items' cross-loading values in the assigned construct are higher than other cross-loading values. Table IV. 4 Cross Loading for Items in Measurement Model Construct Items AF CFP CS CSR ER FP I LR VR Volunntary Responsibility VR1 0.201 0.158 0.261 0.376 0.270 0.106 0.293 0.262 0.817 VR2 0.268 0.246 0.348 0.388 0.381 0.114 0.310 0.326 0.826 VR3 0.296 0.116 0.324 0.359 0.284 0.033 0.292 0.305 0.807 VR4 0.390 0.339 0.498 0.400 0.448 0.187 0.333 0.572 0.648 Legal Responsibility LR1 0.404 0.292 0.462 0.403 0.440 0.243 0.350 0.778 0.440 LR2 0.274 0.270 0.318 0.259 0.312 0.111 0.194 0.624 0.449 LR3 0.126 0.359 0.165 0.342 0.288 0.502 0.239 0.594 0.173 LR4 0.277 0.499 0.380 0.375 0.541 0.221 0.313 0.693 0.257 Economic Responsibility ER1 0.311 0.597 0.423 0.346 0.637 0.263 0.339 0.475 0.261 ER2 0.272 0.506 0.419 0.326 0.677 0.181 0.257 0.562 0.303 ER3 0.513 0.384 0.573 0.659 0.855 0.101 0.592 0.449 0.429 ER4 0.469 0.253 0.449 0.534 0.759 0.125 0.390 0.334 0.280 CSR1 0.520 0.147 0.375 0.602 0.490 0.103 0.419 0.296 0.247 CSR2 0.345 0.327 0.482 0.745 0.536 0.152 0.561 0.486 0.375 51 The results of testing the structural model are used to evaluate the influence between variables in the framework. Before examined the structural model testing Goodness fit (GoF) of the model is needed to determine. In table IV.5, the result of GoF and R 2 values on each endogenous variable were demonstrated. Table IV.5, also present the result of predictive relevance of each endogenous construct (Q 2 ) by performing the blindfolding technique to effectively show how collected data can be reconstructed empirically using the model and PLS parameters (Ali et al., 2016). The Q 2 represents a measure of how well observed values are reconstructed by the model and its parameter estimates, if Q 2 value greater than zero, then it considered to have predictive relevance. The result of Q 2 in this study shows acceptable predictive relevance of perceived behavioural control perceived value, trust and intention to go even though predictive relevance of perceived behavioural control is relatively small. Corporate Social Responsibility CSR3 0.457 0.284 0.521 0.881 0.566 0.069 0.719 0.398 0.403 CSR4 0.422 0.316 0.483 0.819 0.518 0.161 0.623 0.409 0.473 Access to finance AF1 0.759 0.315 0.407 0.397 0.489 0.176 0.541 0.281 0.241 AF2 0.690 0.213 0.331 0.403 0.362 0.184 0.465 0.231 0.159 AF3 0.797 0.334 0.657 0.380 0.392 0.177 0.507 0.382 0.378 AF4 0.784 0.332 0.709 0.512 0.475 0.124 0.559 0.329 0.341 Customer Statisfaction CS1 0.597 0.234 0.780 0.478 0.524 0.048 0.395 0.381 0.489 CS2 0.566 0.413 0.858 0.525 0.590 0.125 0.511 0.356 0.368 CS3 0.519 0.459 0.800 0.492 0.498 0.226 0.524 0.427 0.265 CS4 0.442 0.313 0.492 0.283 0.295 0.208 0.295 0.321 0.298 Innovation I1 0.536 0.434 0.455 0.711 0.482 0.163 0.880 0.364 0.331 I2 0.543 0.436 0.492 0.728 0.458 0.209 0.918 0.358 0.339 I3 0.570 0.386 0.532 0.650 0.506 0.330 0.850 0.355 0.387 I4 0.706 0.385 0.541 0.426 0.528 0.214 0.673 0.317 0.273 Company size, firm age, ownership FP1 0.259 0.067 0.170 0.106 0.101 0.536 0.178 0.252 0.102 FP2 0.121 0.452 0.139 0.128 0.134 0.811 0.211 0.351 0.087 FP3 0.110 0.396 0.140 0.097 0.138 0.757 0.146 0.256 0.129 FP4 0.213 0.282 0.135 0.097 0.180 0.585 0.212 0.225 0.086 Corporate Financial Performance CFP1 0.220 0.598 0.166 0.121 0.207 0.263 0.213 0.142 0.109 CFP2 0.318 0.774 0.330 0.305 0.524 0.409 0.406 0.375 0.239 CFP3 0.310 0.805 0.419 0.260 0.426 0.366 0.357 0.433 0.226 CFP4 0.304 0.839 0.438 0.301 0.402 0.458 0.412 0.505 0.222 52 Table IV. 5 Structural Model Assessment and Goodness-of-Fit (GoF) Index Variable R Square Predictive accuracy Q2 Predictive Relevance Acces to Finance 0.317 Moderate 0.288 Yes Corporate Financial Perfromance 0.458 Moderate 0.242 Yes Customer Satisfaction 0.371 Moderate 0.390 Yes Corporate Social Responsibility 0.520 Moderate 0.470 Yes Company size, firm age, ownership 0.024 Low 0.036 Yes Innovation 0.587 Moderate 0.322 Yes GoF = √������������������∗������ ������ 0.502 Furthermore, the collinearity of structural model was needed to examine. According to Hair et al (2019) collinearity can be measured by Variance Inflation Factor (VIF) value. The ideal score closes to 3 or lower, and if the score higher than 5 means that there is a probability of collinearity among the predictor construct. The structural model in this study is no collinearity, it can be seen in Table IV.6. The results of the hypotheses testing are summarized in Table IV.6. Table IV. 6 Path Coefficient and Hypothesis Testing Hypotheses Label Path Path Coefficient Collinearity β Std. Dev t- statistics p-value Result VIF Result H1 VR → CSR 0.203 0.063 3.206 0.001 Supported 1.373 No Collinearity H2 LR → CSR 0.104 0.098 1.056 0.291 Not Supported 1.698 No Collinearity H3 ER → CSR 0.533 0.089 5.975 0.000 Supported 1.642 No Collinearity H4 CSR → AF 0.563 0.067 8.342 0.000 Supported 1.000 No Collinearity H5 CSR → CS 0.609 0.061 9.955 0.000 Supported 1.000 No Collinearity H6 CSR → I 0.766 0.044 17.377 0.000 Supported 1.000 No Collinearity H7 AF → CFP -0.087 0.102 0.856 0.392 Not Supported 2.575 No Collinearity H8 CS → CFP 0.316 0.084 3.779 0.000 Supported 2.110 No Collinearity H9 I → CFP 0.251 0.092 2.742 0.006 Supported 2.025 No Collinearity 53 H10 CSR → FP 0.157 0.078 2.013 0.044 Supported 1.000 No Collinearity H11 FP →CFP 0.411 0.071 5.775 0.000 Supported 1.082 No Collinearity Significant at: *p < 0.1; **p < 0.05; ***p < 0.01 Figure IV. 1 Path Analysis Result with Inner Outer p-value and Construct R square IV.3 Discussion Based on the findings of the investigation, hypotheses 1, 3, 5, 6, 8, and 9 were accepted, while hypotheses 2 and 7 were rejected. Through enhanced corporate social responsibility (CSR) practises, the implementation of voluntary responsibilities has a positive effect on the financial performance of organisations, according to the findings of the study. This is consistent with previous research conducted in Malaysia, which determined that the establishment of a risk management committee (RMC) is a voluntary practise, but its presence is associated with enhanced financial performance of firms (Boudiab et al., 2021; Ishak & Nor, 2017). Non-profit organisations are required to include information about the source and use of funds in their financial statements, which contributes to the evaluation of their financial performance (Alam et al., 2013; Xie et al., 2019). Non-profit organisations should provide information that reflects their capabilities over a specific period, thereby assisting stakeholders in evaluating their performance (Kaliappen et al., 2019; Thathsarani & Jianguo, 2022).