We explore the bluer star-forming population of the Sloan Digital Sky Survey (SDSS) III/BOSS CMASS DR11 galaxies at \(z>0.55\) to quantify their differences, in terms of redshift-space distortions and large-scale bias, with respect to the luminous red galaxy sample. We perform a qualitative analysis to understand the significance of these differences and whether we can model and reproduce them in mock catalogs. Specifically, we measure galaxy clustering in CMASS on small and intermediate scales (\(r\lesssim 50\,h^{-1}\)Mpc) by computing the two-point correlation function \(-\) both projected and redshift-space \(-\) of these galaxies, and a new statistic, \(\Sigma(\pi)\), able to provide robust information about redshift-space distortions and large-scale bias. We interpret our clustering measurements by adopting a Halo Occupation Distribution (HOD) scheme that maps them onto high-resolution N-body cosmological simulations to produce suitable mock galaxy catalogs. The traditional HOD prescription can be applied to the red and the blue samples, independently, but this approach is unphysical since it allows the same mock galaxies to be either red or blue. To overcome this failure, we modify the standard formulation and infer the red and the blue mock catalogs directly from the full one, so that they are complementary and non-overlapping. This separation is performed by matching the observed CMASS red and blue galaxy fractions and produces reliable and accurate models.