home > faqs > statistics

Statistics for contaminated land

1. Should I apply "statistics" to all sites during contaminated land assessments?
No.  Before, considering applying statistical methods to your data sets the following fundamental questions should be addressed:

• What sampling designs were implemented in the investigation?
• Is there sufficient site data to apply a given statistical method?
• Does the data represent a single population or are there several distributions in the data sets which need to be separated?
• What level of uncertainty relative to the decision making is acceptable?

Answers to the above would help to determine what level of statistics can be applied to the available dataset.
Review all datasets in the context of the Conceptual Site Model before considering using available statistical methods.

2. How should outlier's be treated in a dataset as part of your risk assessment
Considerable caution must be used when removing data, identified as an outlier, from a given data set in order to improve the data distribution.

• "Outliers" in a data set may be an indication of the presence of two or more underlying distributions in the data set or an indication that insufficient data was collected as part of the original site investigation. All data sets should be considered relative the Conceptual Site Model (CSM) of the site, before applying statistical tests;
• There are several available methods for determining outlier's in a data set. In contaminated land scenarios, the Grubb's test, which is often used, is a single outlier test and the presence of multiple outliers may sometimes be difficult to detect. Probability Plots (Q - Q plot) and Histograms are examples of tools which can be used to  quickly visualise the presence of more than one distribution in a dataset. In the context of the CSM, separate the datasets and assess them separately.

3. What tools are available for undertaking statistical tests in contaminated land?
There are several statistical packages and methods (including geostatistical approaches) that can be used for undertaking statistics and assessing data from contaminated land sites.
In May 2008, CLAIRE/CIEH published a useful guidance document "Guidance on Comparing Contamination Data with a Critical Concentration). A commercially available statistical calculator, associated with the guidance has been developed by ESI International. See the link to frequently asked questions on the use of the calculator.
If you have any questions on this, please contact Osric Forton (Osric.forton@ramboll.co.uk) and Phil Studds at Ramboll (phil.studds@ramboll.co.uk)

 

sitemap