The determination of the LSR is still a matter of debate. The classical value of the tangential peculiar motion of the Sun with respect to the LSR was challenged in recent years, claiming a significantly larger value. We show that the RAdial Velocity Experiment (RAVE) sample of dwarf stars is an excellent data set to derive tighter boundary conditions to chemodynamical evolution models of the extended solar neighbourhood. We present an improved Jeans analysis, which allows a better interpretation of the measured kinematics of stellar populations in the Milky Way disc. We propose an improved version of the Stromberg relation with the radial scalelengths as the only unknown. Binning RAVE stars in metallicity reveals a bigger asymmetric drift (corresponding to a smaller radial scalelength) for more metal-rich populations. With the standard assumption of velocity-dispersion independent radial scalelengths in each metallicity bin, we redetermine the LSR. The new Stromberg equation yields a joint LSR value of V_sun=3.06 pm 0.68 km/s, which is even smaller than the classical value based on Hipparcos data. The corresponding radial scalelength increases from 1.6 kpc for the metal-rich bin to 2.9 kpc for the metal-poor bin, with a trend of an even larger scalelength for young metal-poor stars. When adopting the recent Schonrich value of V_sun=12.24 km/s for the LSR, the new Stromberg equation yields much larger individual radial scalelengths of the RAVE subpopulations, which seem unphysical in part. The new Stromberg equation allows a cleaner interpretation of the kinematic data of disc stars in terms of radial scalelengths. Lifting the LSR value by a few km/s compared to the classical value results in strongly increased radial scalelengths with a trend of smaller values for larger velocity dispersions.