By: Sauer S, Lemke J, Buettner R, Kohls N
Publication Name: Personality and Individual Differences
Citation: Volume 81 (2015) 117-123
Whereas the number of studies supporting the efficacy of mindfulness as a health intervention is increasing, the measurement of mindfulness remains a subject of debate. Given the importance of measurement in this field, this paper aims to further our understanding of the assessment of mindfulness by employing an approach referred to as “random forests” (RF). RF is an ensemble learning method that is based on decision trees. RF is well known in biological research, for example, but is practically unknown in psychometrics. In this study, RF was used to gauge the predictive validity of the items from two mindfulness instruments concerning their ability to estimate group allocation (i.e., mindfulness practitioners vs. nonpractitioners). To allow for a better generalization of the results, we examined the research questions in two samples (N = 76 and N = 202) of different quality. We investigated two instruments: the Freiburg Mindfulness Inventory (FMI) and the Mindfulness Attention and Awareness Scale. Although results indicated that both instruments were capable of distinguishing practitioners from nonpractitioners, the predictive quality of most items on both scales was determined to be insufficient.