Duplicates and precision

Paired results from field, coarse and pulp duplicates: fitting a line that bad pairs cannot pull, and measuring precision.

A duplicate is the same sample, or a split of it, analysed twice. The pairs should agree within the precision you expect for that stage, and plotted one against the other they should fall along the 1:1 line. A few bad pairs are normal: a sample mix-up, a nugget. They should not decide where the line goes.

Read

• The Impact of Geological Variation when Quality Controlling Precision (2017): read this first
• A ‘Best Fit’ Line Resistant to Outliers when Assessing Precision (2017): the repeated median
• Identifying Outliers when Assessing Precision (2018)
• A Robust Approach to the Thompson Howarth Chart (2018)

Try

• Repeated median regression workbook: fit a line through your pairs with Siegel’s repeated median, Theil-Sen and least squares, side by side.

In QC Analytical Results

The app plots every duplicate stage for every element and laboratory, with a Theil-Sen line on the scatter and precision charts. How it calculates them:

• RD statistics
• RD Rolling Median
• Thompson-Howarth
• MAD control lines
• Glossary