4.2.1. Multicollinearity

The correlation between independent variables (i.e., the visual appearance of timber joints) with the dependent variable (i.e., overall aesthetic preference) was checked. The preferred bivariate correlation value between the independent and dependent variable is above 0.3 [41]. Collinearity diagnostics were also performed to check further multicollinearity that might not be evident in the correlation matrix. Two values were scrutinised at Tolerance and VIF (Variance inflation factor). The tolerance values for independent variables were above 0.10; the multicollinearity assumption was not violated. These were also supported by the VIF values, as all values are well below the cut-off of 10.

#### 4.2.2. Outliers, Normality, Linearity, Homoscedasticity, Independent of Residuals

Assumptions about outliers, normality, linearity, homoscedasticity, and independent of residuals were verified by inspecting the normal P-P plot of the regression standardised residual and the scatter plot of standardised residuals. In the normal P-P plot, all points are laid in a reasonably straight diagonal line from bottom left to top right, suggesting that there is no major deviation from normality. Additionally, residuals are roughly rectangularly distributed in the scatter plot, with most of the scores concentrated along with the 0 points. The presence of an outlier was checked from the scatter plot (i.e., cases that had standardised residual of more than 3.3 or less than −3.3) and none were found. The presence of outliers was also checked by inspecting the Mahalanobis distance. According to reference [41], the critical value is 18.47 for four independent variables. Three cases above that value were observed, which were slightly above the critical value and hence retained. The maximum value for Cook's distance was also verified to check whether these cases had an undue influence on the result of the model. The maximum value for the Cook's distance was 0.084, suggesting that there was no major problem in the data as reference [42] recommended that a value above 1 would be a potential problem.
