A note on alternative approaches to the multidimensional scaling of similarities data