Reassessing the Cepheid-based distance ladder: implications for the Hubble constant

The Hubble constant ($H_0$) is a key parameter in cosmology, yet its precise value remains contentious due to discrepancies between early- and late-universe measurement methods, a problem known as the "Hubble tension." In this study, we revisit the Cepheid-based distance ladder calibration, focusing on two potential sources of bias in the period-luminosity relation (PLR): (1) the assumed prior for the residual parallax offset of the Milky Way (MW) Cepheids and (2) systematic differences between Cepheid periods in anchor galaxies versus supernova host galaxies. To address the latter, we adopt two different strategies alongside a renewed MW Cepheid calibration. The first strategy involves resampling anchor and host Cepheids from a common distribution of periods. This approach provides a conservative estimate of $H_0 = (72.18 \pm 1.76) \, \mathrm{km/s/Mpc}$, including the renewed MW analysis. The increased uncertainty reflects the reduced sample size -- about 700 Cepheids per resampling compared to 3200 in the original dataset. This method reduces the Hubble tension from $5.4 \, \sigma$ (as reported by the SH0ES collaboration with $H_0 = (73.17 \pm 0.86) \, \mathrm{km/s/Mpc}$) to $2.4 \, \sigma$. The second strategy allows the PLR slope to vary with the period, yielding $H_0 = (72.35 \pm 0.91) \, \mathrm{km/s/Mpc}$, including the renewed MW analysis, and the tension reduced to $4.4 \, \sigma$. A statistical comparison of the model with the single-linear PLR shows a significant preference for the broken PLR (p-value $< 0.001$). Both strategies consistently indicate a downward shift of approximately $-1 \, \mathrm{km/s/Mpc}$ in $H_0$. Our findings underscore the importance of careful consideration of Cepheid population characteristics for precise $H_0$ calibrations.