There is a contradiction between seven meta-analyses, all of which indicate a substantial benefit of the flipped classroom (FC) method for K-12 teaching and some larger study that found no such benefit when compared to “traditional” teaching. In the theoretical part of the paper, we shed light on this contradiction by consulting general literature on meta-analyses. Ranking the 50 included FC studies by the number of classes per experimental condition, we found a negative correlation between the “size” of a study and the effect in favor of FC. In the empirical part, we present an FC study with three conditions concerning mathematical teaching, based on n = 950 students aged 11–13, in which many relevant covariates (e.g., quality of instruction) were addressed. One FC condition was based on students’ knowledge acquisition through instructional videos at home (FCn: n = 12 classes). Considering that self-regulation support might play a crucial role especially for young students working at home, another FC condition (FCS: n = 12 classes) was implemented, in which students could learn additional math-free strategies concerning watching instructional videos. Both FC-conditions were experimentally compared with a control group of traditional teaching (TT: n = 13 classes). No significant effect on learning gains was found between FCn and TT, indicating that “flipping” alone may not be more effective per se. However, a significant difference was found between FCS and FCn. Thus, supporting students’ self-regulation in addition may in indeed open the door to successful FC, even with very young students.