Confidence interval construction for multivariable truncated spline logistic model (MTSLM)

This study highlights the construction of confidence intervals within the Multivariable Truncated Spline Logistic Model (MTSLM). Its truncated spline function allow flexible modeling of nonlinear relationships across sub-intervals that conventional logistic regression cannot capture. A simulation study is conducted to validate the performance of proposed method, and Human Development Index (HDI) data in Indonesia are used as an empirical illustration to demonstrate its practical applicability. The results show that several predictors are statistically significant. For instance, access to clean water has a confidence interval [0.0124, 0.0566], while the percentage of poor population shows interval with range [−0.4039, −0.0341]. Nonlinear effects are evident in the unemployment variable, with both confidence interval [0.5083, 1.2985] and [−1.4760, −0.3656]. Similar patterns are observed for school enrollment rates. The confidence interval analysis result provides insight into the reliability of these predictor effects, emphasizing the role of infrastructure, education, and social-economic factors.The findings demonstrate that MTSLM not only improves accuracy but also provides informative and reliable estimates of predictors influencing HDI.The highlights of this study:•Develop convidence interval for MTSLM to flexibly nonlinear model.•A simulation study and real HDI data are used to validate the proposed method.•The MTSLM model outperforms binary logistic model.