In double-quantitative rough set (DqRS) theory, all kinds of rough set models and attribute reduction methods have recently been examined, and all of them were to acquire knowledge from a particular decision table. However, little attention has been paid to the construction of dominance-based DqRS (DDqRS) models, optimal scale selection and optimal scale reduction in multi-scale dominance intuitionistic fuzzy decision tables (MS-DIFDTs). In reality, this task is very important because it can provide a way to select an appropriate simplified scale for some given objective in MS-DIFDTs. To address this issue, this study mainly focuses on constructing two DDqRS models and selecting the simplest optimal scale of the given MS-DIFDTs. First, a novel ranking approach for ranking IF values is presented and used to construct a dominance relation in IF-valued decision tables. Second, by considering the combination of probabilistic and graded rough set, we propose two types of DDqRSs in MS-DIFDTs. Third, on the basis of the presented DDqRSs, we discuss optimal scale selection and scale reduction approaches to acquire classification or decision rules in MS-DIFDTs. Using the presented approaches, an optimal scale reduct can be acquired in MS-DIFDTs. The results obtained in this study are helpful for users to select a simplest optimal scale for meeting the knowledge representation requirement.