We measure the two-point clustering of spectroscopically confirmed quasars from the final sample of the Baryon Oscillation Spectroscopic Survey ( boss ) on comoving scales of 4 \lesssim s \lesssim 22 ~ { } h ^ { -1 } { Mpc } .
The sample covers 6950 deg ^ { 2 } ( \sim 19 ( h ^ { -1 } Gpc ) ^ { 3 } ) and , over the redshift range 2.2 \leq z \leq 2.8 , contains 55,826 homogeneously selected quasars , which is twice as many as in any similar work .
We deduce b _ { Q } = { 3.54 \pm 0.10 } ; the most precise measurement of quasar bias to date at these redshifts .
This corresponds to a host halo mass of \sim 2 \times 10 ^ { 12 } ~ { } h ^ { -1 } M _ { \odot } with an implied quasar duty cycle of \sim 1 percent .
The real-space projected correlation function is well-fit by a power law of index −2 and correlation length r _ { 0 } = ( 8.12 \pm 0.22 ) h ^ { -1 } Mpc over scales of 4 \lesssim r _ { p } \lesssim 25 ~ { } h ^ { -1 } { Mpc } .
To better study the evolution of quasar clustering at moderate redshift , we extend the redshift range of our study to z \sim 3.4 and measure the bias and correlation length of three subsamples over 2.2 \leq z \leq 3.4 .
We find no significant evolution of r _ { 0 } or bias over this range , implying that the host halo mass of quasars decreases somewhat with increasing redshift .
We find quasar clustering remains similar over a decade in luminosity , contradicting a scenario in which quasar luminosity is monotonically related to halo mass at z \approx 2.5 .
Our results are broadly consistent with previous boss measurements , but they yield more precise constraints based upon a larger and more uniform data set .