In ΛCDM, on sufficiently large scales the matter power spectrum should be statistically homogeneous, so measurements of clustering amplitude and shape in different sky regions or sub-volumes should agree up to sample variance once survey systematics are controlled (Peebles 1980; Scoccimarro 1999). Yet “position-dependent power” and separate-universe style analyses find that the local small-scale power can vary more strongly with large-scale environment, direction, or location than simple ΛCDM plus Gaussian initial conditions and linear bias predict, hinting at either enhanced super-sample effects, non-Gaussian couplings, or unmodeled large-scale structure and observational systematics (Chiang et al. 2014; Barreira et al. 2019).