Meta-d’/d’ has become the quasi-gold standard for measuring metacognitive efficiency in the field of metacognition research because meta-d’/d’ is widely believed to provide control for discrimination performance, discrimination criteria, and confidence criteria. Here, we show that only under one specific generative models of confidence, namely the independent truncated Gaussian model, meta-d’/d’ is guaranteed to control for all of those. Simulations using a variety of different generative models of confidence showed that for most generative models of confidence, there exists at least some combinations of parameters where meta-d’/d’ is affected by discrimination performance, discrimination task criteria, and confidence criteria. These simulations imply that previously reported associations with meta-d’/d’ do not necessarily reflect associations with metacognitive efficiency but can explained by correlations with discrimination performance, discrimination criterion, or confidence criteria. We argue that measuring metacognitive efficiency requires researchers to identify the generative model underlying confidence judgments for each specific task and set of stimuli. For this purpose, we present the R package statConfR, which allows researchers to fit the independent truncated Gaussian model as well as a number of alternative models of decision confidence and/or metacognition. If one of the alternative models of confidence describes empirical confidence judgments better than the independent truncated Gaussian model, we strongly recommend using the metacognition parameter of the corresponding model instead of meta-d’/d’ to quantify metacognition.