Since it’s either impossible or not needed to excavate an entire site, archaeologists employ the use of sampling techniques in order to get a clear picture of an area without having to sort through and analyse extremely large amounts of information and artefacts. There’s also always that pesky problem of funding as well. The harsh reality of the beast is, yeah we can’t get to everything in every region. So we use sampling techniques to make peace with that, and hopefully with the information we get out of the sample, gives us a proper representation of the area.
Now while it sometimes seems lazy or not very productive, I would also like to point out that it’s actually in the interest of best practices to leave areas covered up and untouched for preservation purposes as a site is destroyed as it is excavated. And who knows, maybe in the future, better techniques or technologies will be invented, meaning those covered areas are gonna be well taken care of later on.
Now, the main goal of sampling is to be able to come to sound conclusions of an entire site or region, from a fraction of the area. And the best way to do that is through statistical methods. Most university archaeology programs have a mandatory statistics class for this purpose. I’m sorry my artsy friends, math and science always creep up when you’re trying to run away from them. And to go alongside math, we have theory. Using the statistics, we then employ probability theory. What joy. Much fun. But that’s why we call it: Probabilistic Sampling. Its through the use of math that we make sure the probability of the generalizations we make from this sampling is as correct as possible. It’s the same thing they do when there’s a public opinion poll, or like what they do on Family Feud.
There are four main probabilistic sampling techniques used by archaeologists and they all sound confusing and overwhelming so let’s try and get through it together.
Simple Random Sampling: This is used in order to understand the site as a whole, not just the areas where we know something will be found. Usually the areas to be sampled are chosen at random using a table of random numbers. Nowadays, in order to rule out bias, computer programs are usually used to determine the random sampling areas. Once you define your area, you then determine the size of your sample units, how many units you want and how much of the area you want to sample. Of course, the more you sample, the more accurate of a prediction you will make. It’s all a big balancing act, really.
There are of course, drawbacks to this method. One of course is that you need to define your sire borders beforehand. That means you have to be really sure of the size of your site, and it’s not always so easy to know before the sampling takes place. Secondly, this random assigning of numbers can cause the sampling areas to be grouped in clusters, meaning certain areas will have larger sampling areas than others, making it inherently biased.
Stratified Random Sampling. This is where the region is divided into its own natural zones, such as cultivated land, forests, riverbanks etc. The sampling squares are then chosen by using the same random numbering program as Simple Random Sampling, but in proportion to the size of each natural zone. So for example, if 50% of the region to sample is covered in forest, then 50% of the random sampling zones will be there. This makes sure that the sampling done doesn’t favour one area over another, and everything is proportional to the site.
Systematic sampling, is when you come up with a system for sampling. You essentially make a grid of equally spaced locations to sample. You can say like, every other square, or every two squares. As long as it’s even, it’s up to you! This is kind of the simplest one to follow. The good and the bad about this method is that with such regular spacing, you can either miss or hit every diagnostic thing in an equally regular pattern, which can also cause a bias.
So because systematic sampling isn’t the best option, you can make a hybrid technique!
Stratified unaligned systematic sampling- Or as I call it: SUSS, because you can very effectively SUSS out a situation. This method combines all three of the methods we just talked about. You make a grid, a normal grid with standard sections, then you lay it over the site, orientating it along the main axis of the site. After you have it laid out, you make your own strata a certain regular size. For example, each strata is gonna be a square of 4 squares by 4 squares. Then within that group of 16 squares, one or two squares will then be chosen at random to use for sampling. This method ensure that there is no bias in the sampling, but it also makes sure that the squares are more evenly spaced out along the site.
Have any more questions about archaeology or history? Shoot me an email: firstname.lastname@example.org
Looking to Find Out More?
Archaeological Institute of America- Sampling
The Archaeological Survey: Sampling Strategies and Field Methods
Thomas F. Tartaron, Hesperia Supplements
Vol. 32, Landscape Archaeology in Southern Epirus, Greece 1 (2003), pp. 23-45
The design of archaeological surveys
Michael B. Schiffer, Alan P. Sullivan & Timothy C. Klinger
Pages 1-28 | Published online: 15 Jul 2010
Sampling Strategies- University of Texas
Archaeological Sampling Strategies
Mary Richardson Grand Valley State University
Byron Gajewski The University of Kansas Medical Center
Journal of Statistics Education Volume 11, Number 1 (2003)
Analysis of Archaeological Sampling Methods Using the Complete
Surface Data from the Pirque Alto Site in Cochabamba, Bolivia
UW-L Journal of Undergraduate Research X (2007)
Sampling Strategies and Problems of Archaeological Visibility
P. C. Woodman
Ulster Journal of Archaeology
Third Series, Vol. 44/45 (1981/1982), pp. 179-184