Context. The detection of gravitational waves emitted by binary black holes raises questions regarding the origin of binaries. There are several models in the literature involving binary evolution in both the field and clusters. Aims. We compare the predictions of these models with observations and establish the reliability of this comparison. Methods. We use the likelihood calculation to compare the models in the space spanned by the observed chirp mass and the luminosity distance of the source. Results. We rank the models by their ability to explain all current gravitational wave detections. It is shown that the most probable models correspond to binary evolution with low metallicity. Several variants of such evolution have similar likelihoods. The globular cluster model, considered here, is disfavoured. Conclusions. We present the usefulness of the method in distinguishing between models when new observations become available. We calculate the number of observations required to distinguish between each pair of models. We find that the number varies from ten to several thousand for some pairs of models, yet almost two-thirds of pairs are distinguishable with at most 100 observations.