Archaeologists employ several methods for locating sites within a region. Their choice of method frequently depends on the questions they are asking and how they think they should go about answering them, and on more practical issues such as the amount of funding, personnel, and field time available. Somewhat aypically a n archaeologist might survey the entire area, as has been done in the past in regions such as the Valley of Mexico, to catalog and sample every site. More commonly, an archaeologist will choose to sample a region. Sampling allows one to draw valid conclusions surveying only a small portion of an entire region. The principle is the same that allows political polsters, for intance, to determine to level of public support for this or that candidate or this or that policy. Similarly, sampling (and the entire field of statistical theory that comes with it) allows epidemiologists (specialists in the causes and spread of disease) to study the behavior of disease in human populations, providing timely warning in the event of local and global outbreaks of flu, etc. Archaeologists use sampling to survey only a fraction of a total universe of site, like those in an large geographic basin or arrayed along the ancient course of a river, like the Tigris and Euphrates in ancient Mesopotamia, but still draw conclusions about the relative frequency of this or that type of site (encampment, hamlet, city, quarry, mine,. ...) and aspects of the landscape and regional ecology that they may be associated with. There are two kinds of sampling procedures. One is simple random sampling in which a region is divided into a number of sample cells, perhaps 10 meters in width, perhaps 5 kilometers in width. The archaeologists than chooses some number of cells through a randomizing procedure and surveys those. Humans are inherently biased creatures when it comes to sampling it seems. There is a danger that if not random, the sample might be biased in some unforseen way. A political poster would never just ask his or her office mates their opinion, such a small group, all working in the same profession might share similar ideas and, hence, not be reflective of the entire population at large. There is one famous case where early polsters used the telephone, then still in its infancy. Only a small segment of the population, generally wealthy or professional, had phones. The resulting political poll in no way reflected the mode of the nation, just the mood of those that could afford the then very expensive telephone service. And then only in cities, as the telephone system had yet to expand into rural America. Similar problems beset the archaeologist. Does the non-random sample of sites follow modern trails and roads with no connection to the byways and paths that existed in the period of interest? Did the archaeologist ever get more than a couple of miles away from the nearest source of ice-cold beverage?
In some cases the archaeologist might like to divide the region to be surveyed into a series of subsample zones based on regional ecology or other landscape features of interest. Each one of these zones is then sampled independently of the other. This allows conclusions to be drawn about how each ecozone or other subdivision was utilized by ancient populations. Were the piedmont slopes used for full-time agriculture while the lakeside was used for fishing and reed hrvesting? Did summer bands prefer one ecozone to another? Switching in winter? This type of survey procedure is known as stratified random sampling.
Source: Archaeology Glossary