Articles by Michael J. Baxter
Statistical analysis of some loomweights from Pompeii: a postscript
Michael J. Baxter, Hilary E.M. Cool, Michael Andrew Anderson
Abstract
In 2008 the authors undertook a detailed statistical analysis of the dimensions of a large sample of loomweights from Insula VI.1, Pompeii. A major finding was that the weights of the loomweights appeared to have a bimodal distribution. Further analysis using loomweights that have come to light since the original work strengthens this observation. An analysis of loomweight volume has been undertaken with a view to predicting weight from volume for incomplete loomweights where sufficient information is retained to allow a volume calculation. Recently published experimental work allows an interpretation of the bimodality of the weights in terms of the loom set-up and the cloth being woven, and this is explored in some detail, along with further, tentative, observations on the chronology of the weights.
«Archeologia e Calcolatori» 2010, 21, 185-200; doi: 10.19282/ac.21.2010.09
Correspondence Analysis in R for archaeologists: an educational account
Michael J. Baxter, Hilary E.M. Cool
Abstract
Correspondence Analysis (CA) is a popular tool for archaeological data analysis, appropriate for use with tables of non-negative number. The technique allows the visual display of the associations between the rows and between the columns of a data matrix, and the relationships between them. Archaeologists with this kind of data often have no problem in understanding the ideas behind CA, but with limited training in statistics may have problems in implementing it. Commercial, menu driven, statistical software packages of the type used for service teaching in universities are expensive and restrictive in the way results from a CA can be presented. Archaeologists outside the university sector may not have access to such software. This paper is a guide to how the open-source software R can be used to undertake CA. R is a sophisticated, ‘state-of-the-art’ package that is constantly updated. It is not menu driven and can seem forbidding to new users. The paper provides a detailed account, ranging from installation of the package through to real applications of CA, that has helped, and we hope will continue to help, encourage the use of CA among archaeologists who have previously been discouraged from engaging with it.
«Archeologia e Calcolatori» 2010, 21, 211-228; doi: 10.19282/ac.21.2010.11
Notes on the statistical analysis of some loomweights from Pompeii
Michael J. Baxter, Hilary E.M. Cool
Abstract
Recent work, in the field, on the dimensions and weights of loomweights from excavations in Insula VI.I, Pompeii suggested that there was structure in the form of evidence of bi-modality in the weights. The paper has two purposes. One is to illustrate a variety of statistical methods that were used to confirm the validity of our observations. The other is to discuss what the archaeological implications of this might be. A more general point is that if more attention is given to what are often regarded as “uninteresting” artefacts some interesting results may emerge - specifically, it can be asked whether loomweights have chronological significance for interpreting archaeological sites (at Pompeii at least).
«Archeologia e Calcolatori» 2008, 19, 239-256; doi: 10.19282/ac.19.2008.20
Beyond the histogram. Improved approaches to simple data display in archaeology using kernel density estimates
Michael J. Baxter, Christian C. Beardah
Abstract
The histogram is one of the most widely used descriptive statistical techniques in archaeology, but suffers from a number of well known problems. These include the dependence of its appearance on the choice of origin and interval width. Kernel density estimates provide an alternative to the histogram and avoid some of its problems. They have been little used in archaeology, probably because suitable software has not been widely available. We illustrate some of the advantages of kernel density estimates through several simple examples. Computation was done using the MATLAB package and routines written by the second author that are freely available. One issue in the use of kernel density estimates concerns a choice analogous to that of the interval width for a histogram. Our routines implement several approaches that are discussed in an appendix.
«Archeologia e Calcolatori» 1996, 7, 397-408; doi: 10.19282/ac.7.1996.33
Principal component analysis in archaeometry
Abstract
Principal component analysis (PCA) is a widely used multivariate method in archaeology, and is particularly prevalent in archaeometric applications. The paper reviews the use of the methodology in archaeometry, including the choice of data transformation and standardisation. The related methods of factor analysis and correspondence analysis are also briefly considered. Two detailed examples illustrate some of the methods discussed, including uncommon approaches such as the use of ranked data.
This website uses only technical cookies strictly necessary for its proper functioning. It doesn't perform any profiling and doesn't use third party cookies of any kind.
Read our privacy policy for additional information.
By clicking 'OK' or closing this banner you acknowledge having read this information and accept the website's contents.