In quantitative analysis of highly scattering samples using Vis-NIR spectroscopy, the main problem is that multivariate calibration models built on conventional spectroscopic measurements such as Transmittance or Reflectance are adversely affected by variations arising from non-linear multiple light scattering effects. Because these variations are not necessarily related to changes in the chemical composition, it makes the extraction of chemical information from such samples challenging. Instead of spectral pre-processing, which is commonly used by Vis-NIR spectroscopists to deal with undesirable scattering effects, this work presents an optical methodology to reduce multiple scattering. A new optical setup, based on polarized light spectroscopy is specifically designed to select photons that have been only weakly scattered. In this study, we propose to use a Polarized Light Spectroscopy setup (PoLiS), to optically select photons that have encountered only one interaction with the matter, i.e. photon of which paths have not been affected by multiscattering. We propose to combine the PoLiS spectral information with the Absorption-Remission function defined by Dahm and Dahm in their Representative Layer Theory to compute new absorbance spectra, hereafter referred as Abs_PO, fulfilling Beer-Lambert's law conditions. In order to validate the theoretical principles, PoLiS measurements have been performed in the 400 – 800 nm range on: - Powdered samples mixing sand (i.e. the scatterer) and two coloring dyes (the absorber) of known concentrations - Liquid samples mixing milk and coloring dye. Incident light is linearly polarized and the two components (co-polarized and crossed-polarized) of the diffusely reflected light have been measured by successively rotating over 90° an analyzer placed before the detector. These two components allow computing the classical backscatterd absorbance {-log (R)} and the PoLiS absorbance. These two absorbance spectra have been compared on the basis of (i) the spectral feature analysis and (ii) the linearity between the absorbance value at one wavelength and the concentration of the absorber. The PoLiS method provides significant improvements in the quality of the spectral measurement, showing absorbing features very similar to the real absorbance of the coloring dye diluted in distilled water measured in transmittance with a lab spectrometer. In addition, it has been showed that scattering is responsible of non linearities due to optical interactions and that these unwanted effects are highly reduced with the PoliS method: the PoLiS absorbance spectra happens to be a linear combination of the absorbance spectra of the two coloring dyes. By selecting only the light which has conserved the initial polarization and therefore being less impacted by scattering events, the collected photons all have the same mean free path length. As a consequence, the linear relationship between the spectra and the absorber’s concentration is restored (Pearson’s coefficient at 405 nm is R=0.77 and R = 0.95 respectively for the classical and PoLiS absorbance). The PoLiS method presents a high potential to increase the diffuse reflectance signal quality on highly scattering media, in solid (soils, waste, pharmaceutical tablets) or liquid form (algae, sludge). This optical pretreatment allows us to retrieve linear and steady conditions for spectral analysis.