I would say it is both! Because of the abundance of electrons, the electric field at the battery pole/boundary, at the instant of turning on the switch (t=t0), is quickly (within a few Debye lengths) screened and cannot possibly reach the electrons further down the wire. However, the electrons at the vicinity of the pole that do feel the effect of electric field reposition themselves (pushed/pulled by the force), creating electric dipoles (by partnering up with the positive nuclei), so that the original field can further penetrate into the wire. This reconfiguration of the charges mediates the electric field through the metallic wire and happens very very quickly when the circuit is formed (switch is turned on). So you see, the electrons do push each other, but only through the field they themselves create. It only takes a tiny fraction of the electrons (small polarization density) to establish the field. This is no other than the subject of electric field in a material medium by the way. The only difference being usually they do the time-independent treatment whereas I believe you are asking about the time-dependent aspect of it. The timescale all this happens is $\epsilon/\sigma$ where $\sigma$ is conductivity and $\epsilon$ the permittivity of the medium.