Demographic phenomena show spatial regularity and specific patterns which need to be studied. Development of GIS and Geodatabase brought maps common, and recent developments of spatial analysis allow going further than mapping to measure spatial structure. Neighborhood and scale are two dimensions of demographic data which can’t be no longer ignored. This paper proposes an original exploration of spatial dimensions of demographic variables (density, proportion of 65+in the population, proportion growth) through the systematic analysis of French census data at various scales (6 administrative divisions) and times (6 censuses from 1968 to 2007). Using moran’s I and LISA, we consider the effects of scales and neighborhood on spatial patterns. The results underline the necessity to better take space into account in statistical analysis.