Star formation is a multi-physics, multi-scale process. The multiplicity of physical processes and scales can generate a significant amount of scatter in the outcome of star formation, in particular in terms of key quantities such as the stellar initial mass function (IMF), the star formation rate (SFR), and the star formation efficiency (SFE). By analyzing large data sets that are becoming increasingly available, one can now assess whether variations do actually exist, quantify them, and attempt to explain them via a comparison with (a range of) theoretical models. I will illustrate this by presenting examples related to the IMF in young Galactic stellar clusters and to the star formation scaling relations in nearby galaxies. I will show that current data argues against a universal IMF in Galactic stellar clusters. I will also argue that the star formation scaling relations on galactic scales should include a description of the role of several other physical quantities, instead of being understood in terms of a mere dependence of the SFR on the density of the star forming gas.