We search for a dark matter signal in 11 years of Fermi-LAT gamma-ray data from 27 Milky Way dwarf spheroidal galaxies with spectroscopically measured J-factors. Our analysis includes uncertainties in J-factors and background normalisations and compares results from a Bayesian and a frequentist perspective. We revisit the dwarf spheroidal galaxy Reticulum II, confirming that the purported gamma-ray excess seen in Pass 7 data is much weaker in Pass 8, independently of the statistical approach adopted. We introduce for the first time posterior predictive distributions to quantify the probability of a dark matter detection from another dwarf galaxy given a tentative excess. A global analysis including all 27 dwarfs shows no indication for a signal in nine annihilation channels. We present stringent new Bayesian and frequentist upper limits on the dark matter cross section as a function of dark matter mass. The best-fit dark matter parameters associated with the Galactic Centre excess are excluded by at least 95% confidence level/posterior probability in the frequentist/Bayesian framework in all cases. However, from a Bayesian model comparison perspective, dark matter annihilation within the dwarfs is not strongly disfavoured compared to a background-only model. These results constitute the highest exposure analysis on the most complete sample of dwarfs to date. Posterior samples and likelihood maps from this study are publicly available.