Probability of being a single star for each of the 20 objects considered in this work as evaluated with our framework along with the associated uncertainty (68% and 90% credible intervals). From left to right: Multi-epoch analysis including all available measurements. Each object is marked according to the label assigned using the method described in Sana et al. (2013); single-epoch analysis including only one randomly selected measurement per star; analysis including only two randomly selected epochs (when available).