I understand the OD measurement in a plate reader varies with the volume of the sample measured. Is there a formula to correct for path length without finding the correction factor by myself?
OD600 is a semi-quantitative measure of turbidity and hence bacterial density. Although the Lambert-Beer law doesn't apply to light scattering, the OD600 is a reasonable linear function of bacterial density, if always measured under the same conditions. So just fill all wells with the same volume. You can make a standard curve (OD600 vs bacteria/ml), but that is rarely worth the trouble.