Precision and Accuracy

The spreadsheet of diffusion tube co-location results, made available by the Local Air Quality Management Helpdesk to allow annual mean diffusion tube results to be bias adjusted, also contains information on the precision of the diffusion tubes, in those cases where duplicate or triplicate tubes were exposed. At the request of a number of Local Authorities, the precision data for each laboratory have been brought together in a summary form.

This page contains the following sections. Use the links below to jump to a specific section.

back to main Diffusion Tubes pages

Precision vs Accuracy (Bias)

Precision should not be confused with accuracy. Diffusion tube precision can be described as the ability of a measurement to be consistently reproduced, i.e. how similar the results of duplicate or triplicate tubes are to each other. Accuracy represents the ability of the measurement to represent the ‘true’ value, which, in this case, is defined as the result from the automatic analyser. When averaged over a number of sets of results bias can be evident. This represents the overall tendency of the diffusion tubes to depart from the ‘true’ value, i.e. to systematically over-or under-read when compared against the reference method.

Once identified, bias can be adjusted for to improve the accuracy of diffusion tube results. This is done using bias adjustment factors, which have been found to be specific to a laboratory and tube preparation method.

A spreadsheet database of bias adjustment factors obtained from Local Authority co-location studies has been compiled by the Local Air Quality Management Helpdesk and can be downloaded here.

Unlike bias, poor precision cannot be adjusted for. It can only be improved by careful handling of the tubes in both the laboratory and the field. The two Figures below illustrate the difference between bias and precision. Both sets of results have the same calculated negative bias, shown by the vertical red line, compared with the true value. However, those in the top part of the Figure have poor precision, whereas those in the lower part have good precision (the vertical spread is just a way of displaying the large number of individual results).

Chart of good and bad precision with negative bias

Good vs Poor Precision

For the purposes of Local Air Quality Management, tube precision is separated into two categories, "Good" or "Poor", as follows: tubes are considered to have "good" precision where the coefficient of variation of duplicate or triplicate diffusion tubes for eight or more periods during the year is less than 20%, and the average CV of all monitoring periods is less than 10%. Tubes are considered to have "poor" precision where the CV of four or more periods is greater than 20% and/or the average CV is greater than 10%.

A spreadsheet tool has been developed to calculate the overall precision of a particular co-location study or any sets of duplicate or triplicate results. The tool can be downloaded on the Local Bias Adjustment Factors page.

The distinction between "good" and "poor" precision is an indicator of how well the same measurement can be reproduced. This precision will reflect the laboratory’s performance/consistency in preparing and analysing the tubes, as well as the subsequent handling of the tubes in the field. Any laboratory can show "poor" precision for a particular period/co-location study, if this is due to poor handling of the tubes in the field.

Precision Summary Results

Two versions of the diffusion tube precision summary results are provided; one detailing the full set of recorded precision results for all years and all laboratories, and a second reduced version containing only the last 3 years of recorded precision results for those laboratories that currently provide diffusion tube analysis. The latest precision results can be downloaded below:

Download Summary of Diffusion Tube Precision 2008-2018 (Full) (PDF, 292KB) [Updated March 2019]

Download Summary of Diffusion Tube Precision 2016-2018 (Reduced) (PDF, 82KB) [Updated March 2019]

Please note that the performance of a laboratory may change from one year to another. Therefore, when assessing the performance of a laboratory using the findings in the above Summary, account should be taken of the proportion of "poor" precision co-location results, not just the presence or absence of poor precision co-location results. Given this, particular care should be exercised when interpreting the results for a laboratory with only a few precision results. Some laboratories in the co-location spreadsheet are not represented in the Summary, because there were no duplicate or triplicate co-location results for that laboratory (some co-location studies are carried out using tubes exposed singly).

What to do with Poor Precision Results

Where results show "poor" precision, then they should be treated with caution, and may not be suitable for their intended purpose. If a particular authority has "Poor" precision from most or all of its duplicate or triplicate data sets then it should look at its own tube handling procedures.

If these are judged to be good then it will be appropriate to look at the precision results for its laboratory to see if this may be the explanation. The aim should be to use results from tubes that are giving "good" precision, as this will improve the overall reliability of the annual mean concentrations derived from diffusion tubes.

back to main Diffusion Tubes pages

back to top