ARCHAEOLOGICAL RESEARCH DESIGN
About Research Designs
- a set of instructions or strategies for archaeological problem-solving
- intended to clarify goals and guide procedures of a research project
- emphasis on RD, an outgrowth of processual archaeology (in Culture History archaeology, RD was usually implicit -- determine chronological sequence of region or site)
- RD is now critical to research-based archaeology (no funding without it), and to CRM archaeology, where issues of cost-effectiveness at stake
- in research archaeology: how will you solve the problem?
- in CRM: how will you solve the problem cheaply?
- RD serves 3 functions:
1. delimit research goals and clarify research questions -- ie., a clear statement of the problem
2. outline basic procedures for solving the problem, often through trial formulation, pilot studies
3. minimize error through appropriate measuring, sampling etc.
Research Design from A to Z
- basic elements of RD:
1. statement of the research problem(s); what is the research about?
2. development of a model of the systemic context
3. deduction of testable propositions
4. statement of methods (field and lab)
5. discussion of how results will be disseminated
1. What is Your Problem, Man?
- problem statement orients the research; most obvious but most difficult part of RD
- good problem statement should show theoretical relevence (how the case study links to a larger archaeological issue)
- problem should solveable (in at least one lifetime!); bad problem: what are the effects of the environment on sociocultural change? -- too vague, too open-ended, won=t get funded
- types of research problems (a partial list):
1. cross-cultural studies -- search for cultural regularities by comparing several cultures, using ethnographic and/or archaeological data
- usually relational studies involving two or more variables (eg., degree of sedentism, degree of food storage)
- often geared to dealing with Abig theory@ problems
- inexpensive research, usually does not involve fieldwork
- biggest problem: reliability of existing data drawn from many sources
2. archaeological case studies -- most common, by far; involves intensively examining one case study of a larger problem
- may be used to refute conventional wisdom about the larger problem; eg., Ahunter-gatherers do not have economic specialization@
- biggest problem: demonstating relevence of the case study
3. ethnoarchaeological study -- usually a case study aimed at linking dynamic to static
- problems: is the static worth knowing about (is it an important archaeological pattern)?
- does more than one dynamic produce the same static?
Developing a Model
- model operationalizes theory with repsect to particular research problem; especially relevent to archaeological case studies
- attempts to describe what the cultural system looked like, how it functioned (describes the >dynamic=, but without doing ethnoarchaeology
- modeling is often based on existing anthropological theory (eg., D+D, p. 69)
- in archaeological case studies, modeling may also invoke ethnographic record, using a version of DHA (relational analogy)
- simulation modeling (eg., linear programming) sometimes used; systemic variable states are changed through several iterations
Hypothesis Formulation and Testing
- using deduction, moves research design from description of systemic context to archaeological patterns it produced (eg., D+D, bottom p. 69)
- should also make clear operating assumptions -- conditions that underly the model; we assume their existence without further testing (eg., climatic conditions have not changed in past 5000 years@; Awomen made the pots@)
Measuring the Data
- RD should specify the kind of data appropriate to solving the research problem (ie., what will we make observations on)
- how will data be measured, according to what scale?
- scale possibilities include: nominal (present/absent), ordinal (data can be ranked), interval (data have the property of distance), and ratio (allows comparison of two variables, eg., length:width = 2:1)
- certain research problems often Adictate@ the kind of data required
- eg., Aproblems of association@ (if A (agriculture), then B storage)) usually require nominal scale data; nominal scale places data into categories (present, absent, red, blue, etc.)
- Aproblems of correlation@ (increase in A (distance between sites) leads to increase or decrease in B ceramic stylistic similarity)) require ordinal or interval data
Describing the Data
- an attempt to look for systematic patterns in the data set, once measurements have been made
- two types:
1. description of pattern in the sample of actual observations -- descriptive statistics
2. description of pattern in the larger universe from which sample is drawn -- inferential statistics
- descriptive statistics (mean, median, range, etc.) describe general properties of sample
- inferential statistics reason from sample description (a statistic) to population description (a parameter), using principles of probability; ie., what are the chances that the sample drawn accurately reflects the population?
- inferential statistics are critical to process of induction -- allow us to generalize from specific cases -- producing empirical generalizations (hypothesis formation)
Sampling the Data
- RD should discuss how data will be collected
- different kinds of sampling: judgemental, systematic, random, stratified
- inferential statistics require random sampling (eg., can we infer projectile point length at a site if we only recover from refuse midden?)
Dissemination of Results
- RD should specify how research results will be reported (ie., you must learn to write good)
- what audiences should results be presented to?