Our perception of a blue sky has more to do with Rayleigh scattering than refraction. Due to dipole moments in the molecules that make up the atmosphere, incident light (from the sun) is scattered. The intensity of this scattering is proportional to (1/λ)4, so that shorter wavelengths are scattered more strongly. (Since the eye has cones that detect red, green, and blue light, the net effect is that we perceive the sky to be blue.)
Thanks for the reminder. Can you further detail the dipole mechanism of Rayleigh scattering? (Isn't scattering a type of refraction?)
I guess from the standpoint of photon-electron interaction, reflection, refraction, and scattering are all similar/the same. But yeah, I think that refraction is the result of the combined scattering by the individual molecules of some material. Since these molecules are arranged in some structure, that combination results in the refracted beam and the reflected beam (radiation in other directions destructively interferes). Beyond that, I'm not really qualified to comment; what I said may even be off base.
As far as the scattering in the skies goes, here's what I remember from my undergrad EM course:
we're dealing with a non-relativistic situation where the scattering radius is on the order of the Bohr radius, which << than the wavelength of visible light. The dipoles in the molecules of the atmosphere are driven by the electric field in the incident wave, and the scattered light comes out polarized. So you end up with the Larmor radiation formula with that ω4 dependence. I was going to try to transcirbe my old notes, but this has the same stuff only with better organization.