The carbonate ion plays a central role in the biochemical formation of the shells of aquatic life which is an important path for carbon dioxide sequestration. Given the vital role of carbonate in this and other contexts, it is imperative to develop accurate models for such a high charge density ion. As a divalent ion, carbonate has a strong polarizing effect on surrounding water molecules. This raises the question whether it is possible to describe accurately such systems without including polarization. It has recently been suggested the lack of electronic polarization in non-polarizable water models can be effectively compensated by introducing an electronic dielectric continuum, which is with respect to the forces between atoms equivalent to rescaling the ionic charges. Given how widely non-polarizable models are used to model electrolyte solutions, establishing the experimental validity of this suggestion is imperative. Here we examine a stringent test for such models: a comparison of the difference of the neutron scattering structure factors of K2CO3 vs. KNO3 solutions and that predicted by molecular dynamics simulations for various models of the same systems. We compare standard non-polarizable simulations in SPC/E water to analogous simulations with effective ion charges, as well as simulations in explicitly polarizable POL3 water (which, however, has only about half the experimental polarizability). It is found that the simulation with rescaled charges is in a very good agreement with the experimental data, which is significantly better than for the non-polarizable simulation and even better than for the explicitly polarizable simulation.