It's a little funny that it's usually phrased as "electrons hit atoms"...
Electrons propagate as waves (quantum matter waves), and to them, the crystal looks no different than, if you can imagine an underground cavern, a tight grid of tunnels, effectively an open space with periodic columns supporting the ceiling -- that is the environment around the electrons, and the electrons are waves much as acoustic or electromagnetic waves would propagate through that structure. The columns are the atoms, with high potentials (energy barriers) around them, due to the bound (valence) electrons already "orbiting" the atoms themselves, which exclude free electrons from those areas.
So you can see that electrons, for the most part, simply wash around the atoms. They aren't nearly small enough to "hit" an atom, in the ballistic sense.
Which is possible: that's literally what scanning electron microscopes (SEM) do -- and they need 100keV electrons to do it. The electrons in a solid are bound with only a few eV of energy.
So what does resistance look like? Suppose in that cavern of columns, the columns themselves are jostling about, randomly. Instead of presenting a periodic structure for waves to diffuse (and diffract) through, they will eventually scatter the waves -- the wave fronts partially reflect and diffract around the slightly displaced barriers, and after enough of these subtle interactions, the wave fronts become randomized and diffused. Note that the columns themselves excite waves: just as a stick in a pond creates ripples from its relative motion, the thermal motion of atoms causes ripples in the electron gas in the crystal.
Thermal motion, in turn, is analyzed in terms of phonon waves -- acoustic waves trapped within the crystal. A superposition of these stores thermal energy, and is responsible for the heat capacity of most materials.
The coupling between electron waves and phonon waves is called electron-phonon coupling. It would seem likely that metals with poor coupling have excellent conductivity. I don't remember the magnitude of this effect, if it does vary with resistivity, but it's certainly a contributor.
As it turns out, at room temperature, the thermal motion of electrons is dominant, to the extent that thermal saturation velocity is on the order of 10^5 m/s for electrons in silicon, while drift velocity is on the order of 0.01 m/s!
Note that "drift" is due to the coherent motion of electrons -- not as wavefronts stimulated by an outside current, nothing like that; merely a general trend followed by that sea of rapid movement. Just as an individual water molecule might be zipping around at the speed of sound, but a massive body of water moves cooperatively to exhibit wave effects (whether internal acoustic waves, or gravity waves on the surface).
Coherent movement is observable in fancier semiconductors; in gallium arsenide, the saturation velocity can be exceeded, and ballistic transport occurs. The result is, at a high enough electric field, the conductivity suddenly goes up, which means the equivalent resistance goes down -- negative dynamic resistance! Gunn diodes are not actually diodes at all, but a piece of doped GaAs (no junction at all) which exhibits this behavior. When placed in a resonant cavity, the negative resistance can be used as an oscillator at extremely high frequencies (10s of GHz) -- the effect is atomic scale, so practically unlimited in frequency response*.
*I don't actually know what mechanism(s) limit this. There must be a limit, since as far as I know, this effect hasn't been used for ludicrously wideband amplifiers (1THz+?). The limits due to geometry (frequency is generally inversely proportional to the length scale, due to the speed of light and other equivalent limits within the medium) are orders of magnitude below those of a Gunn diode (i.e., ~100nm versus ~1mm!), so there must be another effect, which I don't know anything about, which prohibits this behavior!
I think the difference is, what type of semiconductor is used; silicon is an indirect bandgap semiconductor, and undergoes avalanche breakdown when critical electric field is applied (such that would give an electron velocity near the thermal saturation velocity). Avalanche you can think of as an electron colliding with an atom to release more (free) electrons and holes; it's often a runaway cascade effect (which gives rise to
switching effects in certain transistors and the noisy voltage of zener diodes). GaAs and many other important semiconductors are direct bandgap, which means they emit light when forward-biased (yay, LEDs!); I don't know if this directly affects avalanche, or if it's purely a bandgap thing (...does cold silicon exhibit ballistic transport?), though.
(Ref: I'm a physicist. At least, within the stuff that I remember.)
Tim