In the laboratory measurement of the sound transmission loss of platelike building elements, a specimen mounted between coupled rooms causes the so-called niche effect. Normally a specimen is mounted inside an aperture in a thick wall, whereas in a special case where a specimen is mounted flush with the wall, a projecting box frame is additionally installed outside the opening. Firstly, the measurements of a glass pane with the two types of niche are numerically modeled by vibro-acoustic simulation, and the niche effect is examined while changing the niche depth and specimen position. As a result, it is revealed that the effect of the projecting niche is generally smaller than that of the recessing niche. Secondly, regarding the cross-sectional shape of the recessing niche, the smaller effect of the staggered niche specified in ISO 10140 is validated by comparison with that of a flat niche. Additionally, the incidence angle dependence of the niche effect is clarified.