Spatial
Data Analysis
in
Archaeology

Anthropology 589b Place: The Scholar's Lab, Alderman
University of Virginia Monday:4:30-7:00
Spring 2007 Fraser D.Neiman


Description

This course covers statistical tools required to discern and interpret effectively spatial patterns in archaeological data at the intra-site and inter-site scales. Specific analytical techniques include kernel density extimation and K-functions for point-pattern analysis, global and local spatial autocorrelation, variograms, interpolation methods, especially kriging, Mantel regression, empirical-Bayes spatial smoothing, unconstrained clustering, correspondence analysis, and spatial regression. We will also consider the design of effective sampling strategies to collect spatial data and links between spatial techniques and the wider orbit of archaeological theory. The course emphasizes the analysis of real archaeological data. Students are encouraged to work on spatial datasets in which they have a special interest. Software applications include SAS, R, and ArcGIS Geostatistical Analyst. Prerequisite: an introductory statistics course.

Course Schedule and Reading

The reading list for the course is available here
Nearly all journal articles and book chapters will be found on Toolkit.
For a list of books on two-hour reserve at Clemons click here

Handouts and Slides
Datasets
SAS Code
  • Autocorr.sas. Computes correlograms and variograms.
  • AutocorrRAN.sas. Computes correlograms, along with 95% confidence limits for Moran's I and Geary's C, based on randomization distributions at each spatial lag.
  • RegressionExamples.sas. Global polynomial and local bivariate fitting.
  • MayaTrendSurface.sas. Global polynomial trivariate fitting.
  • MayaLoess.sas. Local trivariate fitting
  • AR5TrendSurface.sas. Global polynomial trivariate fitting.
  • AR5Loess.sas. Local trivariate fitting
  • LISA.sas. Getis-Ord G, local Moran I, and local Geary C for a specified spatial lag, with randomization-based hypothesis tests. Currently set up for the Maya data.
  • LISA_AR5.sas. Getis-Ord G, local Moran I, and local Geary C for a specified spatial lag, with randomization-based hypothesis tests. Set up for the AR5 house diameter data
  • Mantel.sas.Mantel statistic, with randomization.
  • BetaBinomial.sas.Empirical-Bayes estimation of binomial proportions from multiple samples .
  • CA1.sas.CA using Proc Corresp, and some plots of tyhe results.
  • CA2.sas. Randomization tests of significance for CA axes, and output of data matrix sorted on CA Axis 1 and 2 scores for seriation plotting.
  • MCD.sas. Input data, compute MCDs, and compare to CA scores.
  • SpatialBetaBinomial.sas. Bayesian smoothing of quadrat counts, with CA on smoothed counts
  • KDE.sas. Kernel density estimation, and output to Programita
Problem Sets
Software Tools
  • Introduction to ArcGIS 9.
  • Introduction to SAS
  • SAS. Online documentation. The bits that interest to us are BASE, STAT, and IML.
  • R. Online documentation. Check out the "Manuals" section.
  • GeoDa. From the Spatial Analysis Lab, University of Illinois.

  • Scatterplot Labels for Excel
    An Excel Add-In that allows you to label scatterplots (something that Lotus-123 could do in 1982, but Excel cannot, 20 years later). Download to a folder on your hard drive, double click the self-extracting archive label_97.exe . Checkout readme.doc for further instructions.
  • Frequency-Seriation Diagrams
    An Excel spreadsheet with VBA code by Carl Lipo and Bill Hunt that draws frequency-seriation diagrams a la Jim Ford.
Requirements

Written work for the course includes Problems Sets and a Final Project. Problem Sets will be assigned nearly every week. Completed problem sets will be due the following week, at the beginning of class, when we will discuss them. Your write up should include not only your numerical results, illustrated with appropriate graphics, but also a description of how you got them and what you think they mean, both in statistical and archaeological terms. The text should be no more than three pages, doubled spaced. You should be prepared to offer the class a succinct summary of your work, as a contibution to the class discussion of the problem sets. Late problem sets will not be accepted.

The Final Project is your opportunity to use the methods we have covered in the course to tease substantive meaning out of spatial archaeological data. Note that the emphasis here is on meaning: the project should emphasize analytical methods as tools to make and evaluate defensible inferences about past dynamics from material evidence. You will need to identify an historical or anthropological issue and one or more sets of spatial data that might be used to illuminate it. You are encouraged to choose issues and data in which you have a personal research interest and, therefore, basic familiarity with the current background literature. The data you use should contain information on several different variables that are of potential relevance to your problem.

Over the first several weeks of the course, we will work together to identify suitable issues and datasets for everyone. The topic and dataset(s) for your project must be approved by the instructor by 2.19. During the last class you will give a 20-minute illustrated presentation on your project. Your write up for the final project should not exceed 15 pages (double-spaced), exclusive of graphics. It is due one week after the last class.

Evaluation
Grades of the course will be computed as follows: Problem Sets*.4 + Presentation*.2 + Final Project*.3 + Class Participation*.1.


Last update...01.21.07