Advanced

Computationally intensive multivariate statistics and relative frequency distributions in archaeology (with an application to the Early Epipaleolithic of the Levant)

Stutz, Aaron LU and Estabrook, GF (2004) In Journal of Archaeological Science 31(12). p.1643-1658
Abstract
Archaeologists seek to analyze patterns of similarity and difference among diverse kinds of assemblages that (1) vary in the number of specimens and (2) have been characterized by standard multi-category frequency distributions. Recent developments in computer simulation methods offer marked improvements in our ability to test statistical hypotheses about variation in relative taxonomic or typological abundance data, drawn from assemblages of variable sizes and diverse archaeological contexts (American Antiquity 66 (2001) 715; Journal of Archaeological Science 30 (2003) 837). In this article we extend the highly flexible and powerful computationally intensive framework for statistical analysis to the multivariate family of methods of... (More)
Archaeologists seek to analyze patterns of similarity and difference among diverse kinds of assemblages that (1) vary in the number of specimens and (2) have been characterized by standard multi-category frequency distributions. Recent developments in computer simulation methods offer marked improvements in our ability to test statistical hypotheses about variation in relative taxonomic or typological abundance data, drawn from assemblages of variable sizes and diverse archaeological contexts (American Antiquity 66 (2001) 715; Journal of Archaeological Science 30 (2003) 837). In this article we extend the highly flexible and powerful computationally intensive framework for statistical analysis to the multivariate family of methods of analysis of variation (MANOVA) and non-hierarchical cluster analysis. We treat the relative type-frequency distribution as a multivariate quantitative description of the archaeological assemblage. We then introduce two simulation-based computer applications for analyzing variability between groups of assemblages. We utilize the multivariate applications in a case study; we evaluate how standard microlith typological classifications perform in capturing information about technological and formal variability among Early Epipaleolithic microlith assemblages from the Southern Levant. Computationally intensive, simulation-based statistical techniques allow the researcher to custom-tailor the measure of statistical variation and the model of random archaeological record formation relevant to the given problem. We suggest that with simulation-based approaches, the costs of computer programming and processing are far outweighed by the potential for explaining quantifiable variability in material traces of human activity in the past. (C) 2004 Elsevier Ltd. All rights reserved. (Less)
Please use this url to cite or link to this publication:
author
publishing date
type
Contribution to journal
publication status
published
subject
keywords
multivariate, statistical analysis, relative frequencies, computationally intensive
in
Journal of Archaeological Science
volume
31
issue
12
pages
1643 - 1658
publisher
Academic Press
external identifiers
  • wos:000224838400001
  • scopus:5144223563
ISSN
1095-9238
DOI
10.1016/j.jas.2004.04.005
language
English
LU publication?
no
id
2eb9e38c-f020-4c18-89a5-4ef263262cf8 (old id 262410)
date added to LUP
2007-11-02 10:11:25
date last changed
2017-01-01 04:57:14
@article{2eb9e38c-f020-4c18-89a5-4ef263262cf8,
  abstract     = {Archaeologists seek to analyze patterns of similarity and difference among diverse kinds of assemblages that (1) vary in the number of specimens and (2) have been characterized by standard multi-category frequency distributions. Recent developments in computer simulation methods offer marked improvements in our ability to test statistical hypotheses about variation in relative taxonomic or typological abundance data, drawn from assemblages of variable sizes and diverse archaeological contexts (American Antiquity 66 (2001) 715; Journal of Archaeological Science 30 (2003) 837). In this article we extend the highly flexible and powerful computationally intensive framework for statistical analysis to the multivariate family of methods of analysis of variation (MANOVA) and non-hierarchical cluster analysis. We treat the relative type-frequency distribution as a multivariate quantitative description of the archaeological assemblage. We then introduce two simulation-based computer applications for analyzing variability between groups of assemblages. We utilize the multivariate applications in a case study; we evaluate how standard microlith typological classifications perform in capturing information about technological and formal variability among Early Epipaleolithic microlith assemblages from the Southern Levant. Computationally intensive, simulation-based statistical techniques allow the researcher to custom-tailor the measure of statistical variation and the model of random archaeological record formation relevant to the given problem. We suggest that with simulation-based approaches, the costs of computer programming and processing are far outweighed by the potential for explaining quantifiable variability in material traces of human activity in the past. (C) 2004 Elsevier Ltd. All rights reserved.},
  author       = {Stutz, Aaron and Estabrook, GF},
  issn         = {1095-9238},
  keyword      = {multivariate,statistical analysis,relative frequencies,computationally intensive},
  language     = {eng},
  number       = {12},
  pages        = {1643--1658},
  publisher    = {Academic Press},
  series       = {Journal of Archaeological Science},
  title        = {Computationally intensive multivariate statistics and relative frequency distributions in archaeology (with an application to the Early Epipaleolithic of the Levant)},
  url          = {http://dx.doi.org/10.1016/j.jas.2004.04.005},
  volume       = {31},
  year         = {2004},
}