Representative sampling is a term analytical chemists use as a statistical basis for collecting samples from large sample sources that evaluates the whole source fairly.
This is because when sampling from large areas, it would be possible for the person carrying out the analysis to directly influence the results of the analysis to further a hypothesis as a result from where the samples are taken from. So how can it be used on two-dimensional areas?
How do I know if a sample is representative?
There is no black-and-white answer as to whether a sample is representative or not.
It is more of a scale and you, as an analyst, would want to ideally be towards the representative end of this scale without getting to the point of silliness where you are taking thousands of unneeded samples just for the sake of it.
What is an example of representative sampling?
A field next to a river could be contaminated with extremely high levels of chemical run-off from a production facility located upstream from the field.
To improve the accuracy of the analysis, multiple samples would have to be taken.
However, the distance from the production facility, as well as the distance from the river, would inevitably affect the results of the soil analysis.
What is the next step?
The chemical production facility could hire a team to investigate the field from which the complaint was filed.
They may find the levels of whatever the contaminant may be are well within the safe limits, and no further action would be taken.
However, this method would not work if they took samples that match the hypothesis they desire using unethical practices, such as taking all the samples from the field as far away from the river as possible and/or as downstream as possible.
How do I plan my representative sampling techniques?
When planning your sampling techniques, you must always try to be as unbiased and fair as possible when investigating a claim.
Using the field example, as above, a common method for sampling would be to split the field up into quadrants.
Then you would take a sample from each quadrant and form a map of how the concentration of said contaminant dissipates through-out the field.
The smaller the quadrants, the more representative the sampling.
Is there another sampling method I could use?
Another method for sampling instead of forming this “map” would be random sampling.
This is commonly used when the area is just not feasible to analyse in a quadrant manner due to its immense size.
Where the quadrants are formed, a random number would be selected of its ‘X’ and ‘Y’ axis, taking samples from these points.
It can be quite simple to be representative of a 2-dimensional area. However, when coming to analyse a three-dimensional volume of matter, things often become a little more complex.