The Statistics of Normal Ranges


» Posted by on Mar 5, 2016 in Uncategorized | Comments Off

Normal ranges are being used more and more as generic criteria in environmental monitoring, particularly in Environmental Effects Monitoring (EEM). Testing that an observation falls outside a normal range for a reference data set was described by us in 1998 in our paper (Kilgour et al. 1998) in Ecoscience. Study designs in EEM rarely involve comparison of a single value (exposure value) against a normal range, and more typically involve the case of testing whether a mean of exposure data is > 2 SDs away from the mean of reference data. Our first paper did not deal with this second and more typical case.

With support from Canada’s Oil Sands Innovation Alliance (COSIA), we were finally able to dedicate the time to communicate the steps involved in dealing with the more typical case. So, here is a link to our second paper on the subject (Kilgour et al. 2016). The paper describes the math involved in testing whether a sample (or mean) falls outside a normal range of reference data. There are three ‘approaches’ outlined for each of the sample and mean cases. We used a set of benthic invertebrate community data from the Steepbank River (near Fort McMurray) to illustrate the math.

Here also is the SUPPLEMENTAL INFORMATION, including a Word document with all the math laid out, and an Excel workbook with calculations that you might find helpful.

If you have any questions, please don’t hesitate to contact one of us. We would be happy to help you out.