Objectives. This paper disentangles the degree of concentration and variance in offender residences across different levels of spatial aggregation. Three nested units are analysed simultaneously (and longitudinally) to explore the impact of using different spatial scales, opening prospect for a comparison with existing findings from crime concentration literature. Methods. Data is utilized from West Midlands Police Force, containing the locations of residence for all known offenders in Birmingham between the years 2006 and 2016. Resident locations are aggregated to 3223 Output Areas (OA), nested within 639 Lower Super Output Areas (LSOA), further nested within 132 Middle Super Output Areas (MSOA). Descriptive and model-based statistics are deployed to replicate a recent study for crime events. Results. In contrast to the crime concentration literature, the results indicate that most variance (~50%) in offender residence concentrations is attributable to the largest spatial scale (MSOA level). Output Areas, as the most fine-grained unit, capture approximately 39% of the variance, although this proportion increases during the study period at the expense of MSOA. Conclusions. Findings suggest that the relationship between variance and scale when studying offender residences is not as clear-cut as it is for crimes. Larger units hold some merit (empirically and theoretically), but this is time-dependent, opening up discussions on the role of urban development in determining the appropriateness of spatial scale.