Native PAGE markers contain a cocktail of polypeptides at specific molecular weights and hence they do reflect the apparent weights. But since, charge is also a factor to be considered while performing Native PAGE, how does one account for it?
You don't account for charge no matter what the marker, unless you are using something like blue-native PAGE. Markers in native PAGE are useful for little else than comparing relative migrations across different experiments.
Markers for native gels are not meant for precise determination of protein sizes. As Alejandro pointed out, they allow comparioson of relative migration distances between different gels. Note that basic proteins are usually lost in native gel electrophoreses because they are not able to enter the gel (they move to the opposite direction towards the cathode). The method Adam mentions is a way of getting some size information from markers in native electrophoreses.
It should be noted that in addition to the size and charge of proteins their structure as well as their hydrophobicity have an effect on migration distances in native PAGEs.
I have used the method described by Hedrick and Smith for determination of mol wt in native gel electrophoresis. As mentioned above, it involves running the sample along with known purified protein markers (not denatured) on native gels of different total acrylamide concentrations. As the concentration increases, the gel pore size decreases, and relative mobility changes accordingly. One calculates the change in relative mobility as a function of change in acrylamide concentration for the unknown and also the proteins of known mol wt. The article is by Hedrick JL, Smith AJ., Size and charge isomer separation and estimation of molecular weights of proteins by disc gel electrophoresis.Arch Biochem Biophys. 1968 Jul;126(1):155-64. I found a paper that shows such a mol wt determination in detail. See Fig 2 in this: http://file.scirp.org/Html/3-1350117/1b66db6d-4a4d-4905-8749-8291c9623608.jpg
I have also seen the method applied when the electrophoresis gel buffer was acidic:http://onlinelibrary.wiley.com/doi/10.1111/j.1432-1033.1972.tb01809.x/pdf having run the electrophoresis in acidic gel buffer. So, the method works in both directions. Note that these procedures were carried out when polyacrylamide gels were cast in open-ended glass tubes. I assume, though, that the method would apply to today's slab gel format.