Dependence (influence) relationship among criteria in complex decision-making problems is difficult to identify through supermatrices in the analytic network process (ANP) because of the trouble in preserving consistency. To alleviate the complexity in such situations, the first objective of this paper is to propose a hierarchical ANP that synthesises hierarchical structures and networks, where hierarchies are used to represent outer dependency and networks serve to delineate inner dependence. In the ANP applications, expertise is widely applied to assess influence weights; however, in some cases the criteria assessed are all related to measurable attributes, and it is difficult to estimate their dependence due to the lack of relevant knowledge. Therefore, the second objective of this paper is to introduce a statistical method to obtain influence weights based on measurable data rather than subjective judgement. The third objective of this paper is to present a new technique for deriving global weights, which can largely reduce the computational complexity of powering supermatrices. Finally, a real-world case of Taiwan high-speed rail system is used to demonstrate the use of these approaches.