Detection significance of Baryon Acoustic Oscillations peaks in galaxy and quasar clustering

We compare our analysis of the Baryon Acoustic Oscillations (BAO) feature in the correlation functions of SDSS BOSS DR12 LOWZ and CMASS galaxy samples with the findings of arXiv:1509.06371v2. Using subsets of the data we obtain an empirical estimate of the errors on the correlation functions which are in agreement with the simulated errors of arXiv:1509.06371v2. We find that the significance of BAO detection is the quantity most sensitive to the choice of the fitting range with the CMASS value decreasing from $8.0sigma$ to $5.3sigma$ as the fitting range is reduced. Although our measurements of $D_V(z)$ are in agreement with those of arXiv:1509.06371v2, we note that their CMASS $8.0sigma$ (LOWZ $4.0sigma$) detection significance reduces to $4.7sigma$ ($2.8sigma$) in fits with their diagonal covariance terms only. We extend our BAO analysis to higher redshifts by fitting to the weighted mean of 2QDESp, SDSS DR5 UNIFORM, 2QZ and 2SLAQ quasar correlation functions, obtaining a $7.6%$ measurement compared to $3.9%$ achieved by eBOSS DR14. Unlike for the LRG surveys, the larger error on quasar correlation functions implies a smaller role for nuisance parameters (accounting for scale-dependent clustering) in providing a good fit to the fiducial $Lambda$CDM model. Again using only the error bars of arXiv:1705.06373v2 and ignoring any off-diagonal covariance matrix terms, we find that the eBOSS peak significance reduces from 2.8 to $1.4sigma$. We conclude that for both LRGs and quasars, the reported BAO peak significances from the SDSS surveys depend sensitively on the accuracy of the covariance matrix at large separations.