Where N is the sample size, N’ is the effective samples size, and \(\rho\)  represents the level of autocorrelation. Considering sociodemographic data always contains some level of autocorrelation, Nepal population density does show positive autocorrelation. The Morans' I value for the population density using queen contingency neighbors is 0.29 while the same for log-transformed population density surface has a value 0.58 with a highly significant p-value. Using the spatial auto-correlation parameter is 0.58 and administrative units 35989 in Equation 1 yields 8,166 .  In terms of effective information contained in the dataset, 8,166, judiciously spaced administrative units across the Nepal contain all of the population density information furnished by the 35989 administrative units. This number is valid for whole study area, but if the region is divided, the ESS will vary in different units.  When the study area is divided in three similar regions (ArcMap spatial grouping tool), the Moran's I varies from 0.45,  0.15,  0.53 similarly the sample size varies to 6051, 5138, 2923 total coming up to 14112  units. When the sample size determined on the basis of district level variation the total sum required is 24,226 administrative units. This variation in ESS is due to the  spatial heterogeneity that often exhibited by spatial data. Spatial heterogeneity here refers to uneven distributions or spatial variation of the population across a region.