Tilting the scales: weighing prior dependency and global tensions of CMB lensing

Abstract

We provide a nested sampling analysis of the combination of CMB lensing experiments with other cosmological measurements. Nested samples can be used to compute global consistency statistics between datasets. This is demonstrated for CMB lensing and Baryon Acoustic Oscillations which are uncorrelated, and the correlated case between ACT DR6 and NPIPE lensing. We investigate the effect of the prior widths of the spectral tilt $n_\mathrm s$ used in CMB lensing analyses, which quantitatively, but not qualitatively, affect headline constraints. In the absence of informative priors, SPT-3G performs more similarly to ACT and NPIPE. Bayes factors and the suspiciousness statistic are used to quantify the possibility of tension, and we find the Gaussian assumptions inherent in calculating the suspiciousness tension probability to be unsuitable in the case of strong agreement between CMB lensing experiments.