Element tests, such as water retention tests and triaxial tests, are efficient and widely used methods for investigating the elementary hydro-mechanical behavior of unsaturated soil. In recent research, discrepancies observed in unsaturated test results using axis-translation techniques (ATTs) with ceramic disks and microporous membrane filters (MM filters) have indicated that misunderstandings might be hiding behind the perceived elementary behavior seen in element tests. In this study, triaxial tests with ceramic disks and MM filters were firstly conducted on specimens of unsaturated completely decomposed granite sand, referred to as Masado in Japan, to compare the influence of different techniques on the test results. Then a soil–water-air coupled finite element-finite difference (FE-FD) method, based on a newly proposed unsaturated/saturated constitutive model coupled with a deformation-dependent water retention curve (WRC), was utilized to simulate triaxial tests as boundary value problems (BVPs) because of the non-uniform deformation of the specimens. Drained/vented triaxial tests showed that the stress–strain relation of Masado is basically the same regardless of the applied technique, while the amount of drainage discharge and degree of saturation during the shearing stage are quite different. The calculated average stress–strain relation basically reproduced a tendency similar to that of the tests. It was also shown from the calculation that the ATT with ceramic disks or MM filters significantly influences the distributions of saturation and stress within the specimens during shearing. It should be emphasized that unsaturated triaxial tests with the ATT are basically BVPs instead of so-called element tests.