Percent Obesity

This is a map of the estimated percent of the adult population that is obese within each 250m grid cell. While individual cells may be low or high within close proximity of each other, the overall pattern of obesity is apparent from this map.

Percent Obese vs. National Average: Clusters by Significance

For each 250 meter grid cell we calculated the difference between the percent of population that is obese and the national estimate of the obese population which is 34.9 percent (NHANES, 2011-2012). Then, we ran a spatial cluster analysis on the results to identify areas where obesity is higher than the national average. The color-codes in the map reflect the confidence that the hot spot or cold spot is a statistically significant cluster (at 90% probability, 95% probability, and 99% probability) and not a result of random chance. The analysis technique used is called the Getis Ord-Gi* method which examines each grid cell in the context of neighboring grid cells. The technique identifies clusters by comparing the local sum of values in the neighborhood of each cell as compared to the mean of all cells in the dataset. When the difference in the local neighborhood is unexpectedly different from the overall data, then the neighborhood cells are said to be clustered. The Getis Ord-Gi* technique reports the results using Zscores and calculates the likelihood that the cluster is a true cluster and not the result of random chance. More on the Getis Ord-Gi* technique can be found at: hot-spot-analysis

Percent Obese vs. National Average: Clusters by Z-Score

For each 250 meter grid cell we calculated the difference between the percent of population that is obese and the national estimate of the obese population which is 34.9 percent (NHANES, 2011-2012). Then, we ran a spatial cluster analysis on the results to identify areas where obesity is higher than the national average. As opposed to the 'Clusters by Significance' map, this map is color-coded by the Zscores (not statistical significance) of the analysis. The analysis technique used is called the Getis Ord-Gi* method which examines each grid cell in the context of neighboring grid cells. The technique identifies clusters by comparing the local sum of values in the neighborhood of each cell as compared to the mean of all cells in the dataset. When the difference in the local neighborhood is unexpectedly different from the overall data, then the neighborhood cells are said to be clustered. The Getis Ord-Gi* technique reports the results using Zscores and calculates the likelihood that the cluster is a true cluster and not the result of random chance. More on the Getis Ord-Gi* technique can be found at: hot-spot-analysis