Jump to main content.


Limitations in the 1996 National - Scale Air Toxics Assessment

Information provided for informational purposes onlyNote: EPA no longer updates this information, but it may be useful as a reference or resource.
What are the Results of the Variability Analysis?

The national-scale assessment produced hundreds of millions of estimates of ambient air concentration, exposure and risk. There were estimates of all three of these quantities for each of the air toxics compounds and for each of more than 61,000 census tracts. This can be an overwhelming amount of information, and so it has been organized here to address some of the most important questions that might be asked about variability of these quantities across geographic regions. This organization is into three levels of detail:

County level. The results from separate census tracts in a county have been combined to provide estimates of ambient air concentration and exposure in each county (risks are considered only at the national scale discussed below). This was done by taking the median ambient air concentration and average exposure for all census tracts in a county. This provides an understanding of how these average quantities vary between counties. The maps that have been produced allow the reader to locate a specific county; to determine the ambient air concentration and exposure in that county; and to make comparisons with other counties in that State or in the Nation. In these maps, it is not possible to determine the variation across different census tracts in the county, since only a median (for air concentrations and exposure)for the county is given. As discussed in the section on Limitations, this is a deliberate feature of the study. It was not felt that the accuracy of the study methods allowed a direct comparison of these quantities at the level of individual census tracts, but that it did allow a comparison between counties since the calculation of these quantities in a county would "average out" some of the inaccuracies that exist at the level of census tracts.

As an example, consider the map produced for the variation of the ambient air concentration of benzene throughout the counties of North Carolina. To reach this map, click here and then select Benzene and North Carolina in the options boxes. The resulting map shows the typical ambient air concentration for benzene in each county, given as the median over all census tracts in that county. It may be noted that the counties near the eastern coast generally have lower values of this quantity (below 0.65 micrograms per cubic meter) than do counties near the middle of that state (where the values are between 0.81 and 4.75 micrograms per cubic meter).

Separate maps are available to view the variability of ambient air concentration, exposure and risk across the counties in each state. For both the exposure and risk sites, there are separate maps showing the variation in values for a typical individual.

State level. The results from separate census tracts have been combined to provide estimates of the variation in ambient air concentration and exposure between different census tracts in each state of the study area (risks are considered only at the national scale discussed below). These results are presented as bar charts showing the range of each value in each state. For example, the charts for ambient air concentration for an air toxics compound (such as benzene) and a state (such as Alabama) display the 25th percentile, the median and the 75th percentiles. These represent the air concentration for which it may be said that 25% of the census tracts in that state have an air concentration below this value; and so on for the median and 75th percentiles.

As an example, consider the chart produced for the variation of the exposure to benzene throughout the census tracts of North Carolina. To reach this map, click here and then select Benzene in the options box for Statewide Estimates. The resulting chart shows the variation in typical exposures to benzene in census tracts throughout the state. It may be noted that the 25th percentile is slightly below 0.7 micrograms per cubic meter; the median is slightly below 1 microgram per cubic meter; and the 75th percentile is approximately 1.5 micrograms per cubic meter. This may be interpreted as meaning that 25% of the census tracts in North Carolina are characterized by a typical level of exposure below 0.7 micrograms per cubic meter. Similar comments may be made for the median and 75th percentile.

Separate charts are available to view the variability of ambient air concentration and exposure across the census tracts in each state. For both the exposure and risk sites, there are separate maps showing the variation in values for a typical individual.

National level. The results from separate census tracts have been combined to provide estimates of the variation in ambient air concentration, exposure and risk between different census tracts in the entire study area. These results are presented as bar charts showing the range of each value in the Nation for each air toxics compound. For example, the charts for lifetime cancer risk for an air toxics compound (such as benzene) display the 5th percentile, 25th percentile, median (50th percentile), 75th percentile, 95th percentile and 99th percentile. These mean, respectively, the cancer risk for which it may be said that 5% of the census tracts in the entire study area are characterized by a typical excess lifetime probability of cancer below this value; and so on for the other percentiles.

As an example, consider the chart produced for the variation of the cancer risk from exposure to benzene throughout the census tracts of the entire study region (i.e. the Nation). To reach this chart, click here and then select All in the options box for Cancer Risks Associated With… (which will show the risks from all sources of benzene combined). The resulting chart shows the variation in typical cancer risks from exposure to benzene in census tracts throughout the Nation. It may be noted that the 5th percentile is at a cancer risk of approximately 5 in a million; the median (50th percentile) is slightly below 10 in a million; and so on. This may be interpreted as meaning that 5% of the census tracts in the entire study region are characterized by a typical cancer risk of approximately 5 in a million or less. Analogous comments may be made for the 25th, 50th, 75th, 95th and 99th percentiles using this chart. For example, 95% of the census tracts are characterized by an upper end cancer risk of approximately 50 in a million or less, meaning that 5% of the census tracts are characterized by an upper end cancer risk of approximately 50 in a million or greater. Similar bar charts are available on the same page for noncancer risks and for cumulative risk (i.e. the risk from all air toxics compounds combined).

Separate charts are available to view the variability of ambient air concentration, exposure and risk across the census tracts in the entire study area (the Nation). These charts are the same as those described in the section on State Level above; simply examine the "National" bars in those charts rather than the bar for an individual state. For both the exposure and risk sites, there are separate maps showing the variation in values for a typical individual.

What are the components of variability?
Which components of variability did the national-scale assessment include?
How was the variability analysis conducted?
How can these results be interpreted?

Return to Main Variability Page


Local Navigation


Jump to main content.