ABSTRACT
In Chapter 2, we introduced the crosswise model which can be regarded as a non-randomized version of the randomized Warner model. Despite the advantages we have described for non-randomized models, the crosswise model still inherits the low efficiency from the randomized Warner model. To improve the relative efficiency, in this chapter we systematically introduce a triangular model, which can be viewed as a variant of the crosswise model (see Yu, Tian & Tang, 2008; Tan, Tian & Tang, 2009; Tang et al., 2012).
