Ever since the early days of archaeology, informal criteria or personal judgment has been used as a method for gathering data samples (nonprobabilistic sampling). In this approach, samples are selected by informal criteria such as prominence and accessibility. This usually means that data is acquired from the most obvious and/or most readily available archaeological remains. That means in many cases the archaeologists are guided by precedent, intuition, or guesswork. This is not to say that such uncontrolled sampling methods are wrong; no archaeologist would ignore prominent or obvious remains, or deny their ability to locate archaeological evidence, whether on the basis of experience, intuition, or some ‘sixth sense.’ It is also true, however, that nonprobabilistic sampling has often been accompanied by a disregard for defining the population or the data universe. Unless these definitions are precisely made, and unless the researcher can state in what ways the defined entities are related, no one can judge the likelihood that the acquired sample is representative of the overall population and universe.
Under normal statistical testing, a sample distribution, rather than the parent population, is used to determine the statistical probability or significance, then a decision made to accept or reject the hypothesis based on the results. If we reject or accept a hypothesis that is based only on a sample population, we will always have to admit the possibility of error from an unrepresentative sample. When examining Greek Doric Temples, however, it is the parent population, more or less, that is being tested since the overall population of extant Doric temples is, in fact, quite small. The major advantage of statistical procedures over intuitive methods is the knowledge that it will significantly reduce the risk of error. Probability theory enables us to evaluate the risk and take them into consideration. Thus, deciding upon the criteria used for accepting or rejecting the hypothesis, one expects to come away with some level of confidence as to Doric temple predictability, at least with regards to their proportionality.
A sample universe [Binford 1964] large enough to accommodate the entire population of Doric temples, as defined by CLASS in the established taxonomy, is employed for this investigation. This sample universe was drawn from a non-arbitrary sample unit that covers Greece, Asia Minor, Sicily, and Italy. The sampling frame consists of most known extant CLASS ‘D’ temples and Order ‘e’ structures within the sample universe. The following is a list of these temples used in the sample broken down by type: