Table 4.

Assumption tests for multinomial logistic regression.

Assumption	Test/Method	Result	Interpretation
Independence of Irrelevant Alternatives (IIA)	Hausman-McFadden test	Not violated (p > 0.05)	The choice between categories is unaffected by the inclusion of other alternatives.
No Multicollinearity	Variance Inflation Factors	All VIFs <4, no multicollinearity detected	Predictors are not highly correlated, ensuring reliable estimates.
Linearity in the Logit	Box-Tidwell test	No significant interactions (p > 0.05)	The relationship between predictors and the logit is linear.
Goodness-of-Fit	Deviance and Pearson Chi-Square	Model fits well (p > 0.05)	The model adequately fits the observed data.
Overall Model Significance	Likelihood ratio test	Significant improvement (p < 0.001)	The predictors as a group significantly improve the model fit.