The current work builds on research demonstrating the effectiveness of Productive Failure (PF) for learning. While the effectiveness of PF has been demonstrated for STEM learning, it has not yet been investigated whether PF is also beneficial for learning in non-STEM domains. Given this need to test PF for learning in domains other than mathematics or science, and the assumption that features embodied in a PF design are domain-independent, we investigated the effect of PF on learning social science research methods. We conducted two quasi-experimental studies with 212 and 152 10th graders. Following the paradigm of typical PF studies, we implemented two conditions: PF, in which students try to solve a complex problem prior to instruction, and Direct Instruction (DI), in which students first receive instruction followed by problem solving. In PF, students usually learn from their failure. Failing to solve a complex problem is assumed to prepare students for deeper learning from subsequent instruction. In DI, students usually learn through practice. Practicing and applying a given problem-solving procedure is assumed to help students to learn from previous instruction. In contrast to several studies demonstrating beneficial effects of PF on learning mathematics and science, in the present two studies, PF students did not outperform DI students on learning social science research methods. Thus, the findings did not replicate the PF effect on learning in a non-STEM domain. The results are discussed in light of mechanisms assumed to underlie the benefits of PF.