Figure 4.

Circular buffer sizes associated with nonsignficant bias in effect estimates due to point displacement across multiple spatial smoothness levels of ancillary continuous raster data sets. Overall, use of urban buffer sizes between 1 - 5 km, and rural buffers between 1 - 10 km resulted in unbiased effect estimates across surfaces with moderate to high spatial autocorrelation (ρ ≥ 0.5). Under surfaces of low spatial autocorrelation, no neighborhood combinations yielded unbiased effect size estimates.