But to be a bit more helpful, there are two completely different rationalizations for Raman spectroscopy (which is something I always found nifty). One is the 'semi-classical' way of looking at it. View light as a particle (photons) and Raman spectroscopy is the inelastic scattering of photons by molecules. So the photons that are coming out will differ in energy from the ones going in by certain amounts, depending on how much energy they transferred (or received) from the molecule. That, in turn, tells you about the existing energy levels of the molecule.
The other way to view it, is that the photon is absorbed, corresponding to an excitation to a virtual, i.e. non-existent, energy level. Which does not violate quantum physics if the excitation is for a short enough time (c.f. the time-energy uncertainty principle). So the photon is rapidly re-emitted, corresponding to a transition back to a higher or lower real energy level.
One reason Raman spectroscopy is useful is that it has different, often (always? I can't recall) complementary, selection rules to IR spectroscopy (which is limited to dipole-moment transitions). So transitions that are weak/'prohibited' in the IR spectrum are strong in the Raman spectrum, and vice-versa.