Statistical Methods in Archaeology

Statistical methods and reasoning in archaeological research: a review of praxis and promise

by Dwight Read

Published in Journal of Quantitative Anthropology 1 (1989) 5 - 78.

Evaluating the Distributional Approach to Inferring Marketplace Exchange

by Christopher Garraty

Garraty, Christopher P. (2009)

Over a decade ago Kenneth Hirth (1998, 2000) developed a “distributional approach” for archaeologically inferring the... more Over a decade ago Kenneth Hirth (1998, 2000) developed a “distributional approach” for archaeologically inferring the existence of marketplace exchange based on analyses of domestic artifact collections. Domestic collections, he reasoned, will be relatively homogeneous in areas where most or all households rely on marketplace exchange to acquire domestic provisions. The present study evaluates Hirth’s distributional approach using a statistical measure of diversity (heterogeneity) to quantify variability among domestic collection units over a large area. The data for this study come from theMiddle Postclassic lower Blanco region of Veracruz (A.D. 1200–A.D. 1350), an unknown context of marketplace exchange. A comparison of diversity scores calculated on surface sherd collections from the lower Blanco region with scores from Late Postclassic Teotihuacan (A.D. 1350–A.D. 1520)—a known context of marketplace exchange—suggests the existence of a marketplace exchange system in the lower Blanco region, likely centered at the town of El Sauce. In addition, changes in intercollection diversity (sherds) and obsidian concentrations with increasing distance from the center suggest El Sauce’s market service area encompassed a radius of approximately six to nine kilometers.

The Politics of Commerce: Aztec Pottery Production and Exchange In the Basin of Mexico, AD 1200--1650

by Christopher Garraty

2006 The Politics of Commerce: Aztec Pottery Production and Exchange in the Basin of Mexico, A.D. 1200-1650. Ph.D. Dissertation, Arizona State University, Tempe. Ann Arbor: University Microfilms.

The relationships between market and political institutions have varied in different times and places, but no market... more The relationships between market and political institutions have varied in different times and places, but no market system was (or is) devoid of political involvement. The contrasting approaches of the Aztec empire and Spanish colonial regime to the Basin of Mexico market system are instructive about the ways that commercial agents (producers, traders) respond to “top-down” pressures from state elites to steer and direct the commercial economy to their political advantage. The results of this study suggest that the market system in the Basin flourished under the Aztec empire but suffered a decline after the Spanish conquest. To establish a window on state-market relationships, I focus on pottery production and exchange (plainware and decorated wares) prior to and during the period of Aztec imperial rule (ca. A.D. 1200-1520) and subsequent colonial period (ca. A.D. 1520-1650) based on compositional analyses and analyses of form specialization and attribute standardization. In the fragmented political landscape that preceded the Aztec empire, most plainware producers manufactured on a relatively small scale and exchanged their wares locally through a system of small, non-hierarchical market networks that likely operated independently of elite regulation. Conversely, decorated Black-on-orange and redware serving vessels were manufactured on a larger scale in fewer production loci and exchanged over a wider area, indicating a hierarchical exchange system that operated under elite auspices. During the Aztec empire, the consolidation of power under the imperial capitals of Tenochtitlan and Texcoco brought about a more stable milieu for inter-polity interaction. In this context, plainware and Black-on-orange production both involved large-scale, high-intensity production industries centered at or near four principal market centers in the Basin, including the imperial capitals. Tenochtitlan became by far the most prominent and prolific locus of pottery production and export, especially for Black-on-orange vessels and comales (tortilla griddles). After the Spanish conquest, the large-scale pottery production and export industries evident in the Late Aztec period collapsed. Production was generally less intensive, smaller in scale, and probably mostly geared toward local consumers. Tenochtitlan—now Mexico City—was no longer the principal hub of indigenous commerce and became increasingly geared toward the Spanish overseas economy.

Imperial and Social Relations In Postclassic South-Central Veracruz, Mexico

by Christopher Garraty

Garraty, Christopher P., and Barbara L. Stark (2002)

We explore social and imperial relations in the western lower Papaloapan Basin, especially along the lower Blanco... more We explore social and imperial relations in the western lower Papaloapan Basin, especially along the lower Blanco River, using statistical analyses of ceramic rims from recent surveys. This region is sandwiched between two known tributary provincial centers of the Aztec empire, but its relationship to the empire is uncertain in colonial documentary materials. Our analyses illuminate changes in social relations from the Middle (A.D. 1150-1350) to Late Postclassic (A.D. 1350-1520) periods and shed light on the impact of Aztec imperialism. We use a ceramic unmixing procedure to assign collections to the Middle and Late Postclassic periods for assessment of settlement patterns. Next we use cluster analyses to examine vertical wealth and status differentiation. In the Middle Postclassic period, we observe a concentric gradation of wealth and status away from the small center of El Sauce. Late Postclassic changes include the decline of El Sauce and the founding of a new center at Callejo'n del Horno. The concentric model does not apply to the Late Postclassic period, however, and wealth and status became more highly concentrated at Callejon del Horno compared to its hinterland. We also investigate sparse collections-those with few Postclassic rims-to evaluate whether these collections represent poor residences or, rather, sherd scatter from possible field manuring. The lowerBlanco region was likely integrated into the Aztec empire on the basis of changes in vertical social differentiation from Middle to Late Postclassic times and percentages of Aztec-style ceramics compared to known Aztec provincial centers, especially Cotaxtla.

Böhmische Dörfer? Zur Stellung der Dresdener Elbtalweitung zwischen sächsischer und böhmischer Bandkeramik. In: T. Doppler/B. Ramminger/D. Schimmelpfennig (Hrsg.), Grenzen und Grenzräume? Beispiele aus Neolithikum und Bronzezeit. Fokus Jungsteinzeit – Berichte der AG Neolithikum 2 (Kerpen-Loogh 2011) 11–24.

by Thomas Link

At the periphery of Linear Pottery culture (LBK) dispersal in Saxony, the Dresden Elbe valley occupies a spatially... more At the periphery of Linear Pottery culture (LBK) dispersal in Saxony, the Dresden Elbe valley occupies a spatially advanced island position towards the southeast. At the same time, this location plays an important role, providing a geographical link to Bohemia. Within the wide river valley a spatial pattern of settlement dispersal can be observed, one large and dense concentration of sites, and two smaller agglomerations of settlements. Moreover, the distribution of Early Neolithic sites is confi ned to the left (southern) river bank with its loess soils. The Elbe River, or rather the loess distribution which runs more or less parallel to it, also denotes a north-eastern border of LBK dispersal beyond the Dresden region.

The LBK tradition in Saxony is not a homogeneous entity but comprises three separate regional groups. Although the territories assigned to these groups are separated by only short distances, density mapping indicates the existence of distinct boundaries. The cultural relevance of these boundaries is reflected, for example, in a distinct change in house orientation between the area of the Dresden Elbe valley and Central Saxony, and also in the spatial distribution of ceramic features. On the basis of correspondence analyses of pottery decoration, an intermediary position of the Dresden valley between the Saxon and Bohemian Linear Pottery culture is discerned. Furthermore, it is apparent that the orientation and the intensity of interregional relations are subject to temporal changes, with connections with Bohemia increasing at the end of the LBK period.

Die frühesten Metalldolche Südost-und Mitteleuropas (Praehistorische Zeitschrift 68, 1993/1, 103–145) and Die früheste dolche Bulgariens (Anatolica XVIII, 1992, 61-69)

by Ivan Vajsov

This article is of the earliest metal daggers, which in some cases are hardly distinguishable from knives, is compiled... more This article is of the earliest metal daggers, which in some cases are hardly distinguishable from knives, is compiled in the area between southern Bulgaria (Rohdope mountains), the western Ukraine and Upper Austria (Mondsee). They can be dated to the so called transitional phase between the Eneolithic and the Early Bronze Age according to current Bulgarian terminology. The oldest examples can be placed in the earlier part of the 4th millennium B. C. and are associated with the Bodrogkeresztur Culture. An analysis of the formal aspects of these daggers allows them to be divided into 12 types (with variants), which only partially show genetic connections. The distributions of the individual dagger types do not necessarily correspond to the areas occupied by archaeological cultures defined by pottery. Groupings based on spectral analysis result two main geographic concentrations. Daggers made with arsenic alloys are found in the western Pontic area where as west Carpathian and central Danubian examples show no arsenic inclusions. Interestingly, some Bodrogkeresztur type daggers in Transylvania are made with arsenic alloys, others are not (Arıuşd, Mereşti and Mastacän). Therefore we assume that the daggers first spread from west to east and later dispersed in the opposite direction. The earliest dagger horizon dealt with here, is seen as being a fundamental part of the deep seated changes which lead to the formation of the Early Bronze Age after Aegean terminology.

Das Grab 982 und die Protobronzezeit in Bulgarien

by Ivan Vajsov

This article is a little addition to the study of the earliest metal daggers "Die frühesten Metalldolche Südost-und Mitteleuropas. - Praehistorische Zeitschrift 68, 1993/1, 103-145." It is addressed mainly to the dagger from the tomb № 982 of the prehistoric necropolis in Durankulak (Northeastern Bulgaria). Investigations of this dagger is of great importance to the prehistory of Europe. The period in which it occurs is associated with the collapse of one of the most powerful cultural system in Europe, that of late Eneolithic cultures in Bulgaria (about 4200 years BC). Long hiatul chronologically in the second half of IV millennium BC Dobrogea (Northeastern Bulgaria) is again inhabited by tribes coming from Nord Pontic steppes. These people are already aware of an entirely new, progressive method for enrichment of copper ores with a mixture of arsenic (As). This is a completely new technology for Europe and is essential to the Bronze Age.

Discovered in a tomb 982 from Durankulak (Northeastern Bulgaria) is discussed in the broader context in reflecting the turbulent historical processes in the second half of IV millennium BC. Arsenic-bronze (As) of which the earliest daggers and axes of the fourth millennium BC, have the same production technology. This progressive technology happens after "hunger for metal," typical of Europe in IV millennium BC and a new "know-how", reached Europe through the migration of northern nomads. The introduction of this technological revolution, which eventually paved the way for the emergence of classical bronze metallurgy.

Landeserschließung des Thüringer Beckens in der Merowingerzeit hinsichtlich der naturräumlichen Standortwahl sowie der Bodenverhältnisse

by Armin Volkmann

Merowingerzeitliche Millefioriperlen in Mitteleuropa

by Armin Volkmann

Ceramics From the ‘Sutny’ LBK Settlement at Těšetice-Kyjovice, Moravia, Czech Republic: Processing and Statistical Analyses

by Ivana Vostrovská

Co-authored with Lubomír Prokeš

The aim of this article is to demonstrate the use of multivariate analysis of ceramics to verify the chronology of the... more The aim of this article is to demonstrate the use of multivariate analysis of ceramics to verify the chronology of the ‘Sutny’ Linearbandkeramik culture (LBK) settlement at Těšetice-Kyjovice, in the Znojmo district of Moravia, Czech Republic. In this articlewe present the results obtained from the ceramics that have been processed to date. Based on this analysis we dated the settlement toapproximately the Ib-IIa phase of the Moravian LBK.
A new method for processing Moravian LBK ceramics is suggested, along with a proposed descriptive system, as a formalized descriptionis very important to subsequent statistical evaluation.
We introduce several statistical analyses suitable for the evaluation of ceramic assemblages and demonstrate the application of othermethods of analysis and possible alternatives for archaeological data visualization. For this purpose we used the’R’ open statisticalsoftware package. We hope that this article will contribute to the development of a unified method of processing Moravian LBK ceramics,demonstrate new possibilities for evaluating archaeological data, and simplify its interpretation.

P. Verhagen (2009). Predictive models put to the test. In H. Kamermans, M. van Leusen & Ph. Verhagen (eds.): Archaeological prediction and risk assessment. Alternatives to current practice. Leiden University Press, Leiden (ASLU 17), pp. 71-122.

by Philip Verhagen

Available as chapter 7 of my thesis

In 2002, the research project ‘Strategic research into, and development of best practice for, predictive modelling on... more In 2002, the research project ‘Strategic research into, and development of best practice for, predictive modelling on behalf of Dutch cultural resource management’ (Kamermans et al., 2005; van Leusen et al., 2005) started out by identifying the research themes that were considered to be of great importance for the improvement of the quality of predictive modelling in the Netherlands . One of these themes concerned testing of currently used predictive models, as a means to assess their quality. Very little seemed to be known about the best way to test a predictive model, and in practice tests that have been carried out were limited, and have not been used in a systematic manner to improve the predictive models. At the same time, more and more data sets have become available for predictive model testing because of the enormous increase of archaeological research carried out in the Netherlands, following the ratification of the Valletta Convention. It was therefore decided that the subject should be studied in more detail in the second phase of the project. The current chapter is the result of this more detailed investigation, which has been carried out between January and July 2005.
The research questions defined for this study are:
- Can we identify testing methods that can measure predictive model quality in an unambiguous way, and that will allow us to say whether model A is doing better than model B?
- If these methods exist, what kind of predictive models do we need in order to apply them?
- Is testing really necessary? Can we perhaps deal with the issue of predictive model quality by defining statistical probabilities and confidence limits?
- And finally: do we have the data sets that will allow us to carry out predictive model testing in a rigorous way? And if not, how can we generate these in the future?

These questions are addressed through a review of existing testing methods for archaeological predictive models, that have appeared in and outside archaeological literature since the late 1980s. This analysis is followed by an exploration of the potential of currently available archaeological data sets in the Netherlands for predictive model testing purposes. In teh next section , the question of suitable models will be addressed: what testing methods are applicable to different types of models, and can we identify the model types best suited for testing? In the last section, the conclusions and recommendations of the study will be presented.

P. Verhagen (2006). 'Quantifying the qualified: the use of multi-criteria methods and Bayesian statistics for the development of archaeological predictive models' in: Mehrer, M. & K. Wescott (eds.), GIS and Archaeological Site Location Modeling. CRC Press, Boca Raton, pp. 191-216.

by Philip Verhagen

Available as chapter 4 of my thesis

Over the past ten years, archaeological predictive modeling in the Netherlands has been the subject of a... more Over the past ten years, archaeological predictive modeling in the Netherlands has been the subject of a sometimes-heated debate (see Verhagen et al., 2000). After seminal publications by Wansleeben (1988), Ankum and Groenewoudt (1990), and Soonius and Ankum (1990), a number of predictive maps have been produced by public archaeological institutions in the Netherlands (RAAP and ROB ). At the same time, academic archaeologists have studied the methodological and theoretical aspects of predictive modeling, and criticized the modeling concepts used in public archaeology in several publications (Wansleeben and Verhart, 1992; van Leusen, 1993; 1995; 1996; Kamermans and Rensink, 1999).
The inferential or inductive approach, already criticized in Brandt et al. (1992) for its inability to cope with the low quality of many archaeological datasets, has gradually been replaced by a more intuitive way of model development, trying to make the best of both worlds by including quantitative data when they are available. Coupled to this development towards more deductive mapping, it is notable that the multi-variate approach has been replaced by the use of a reduced number of variables, that are supposed to have the strongest predictive power for a particular region or archaeological period. These models can best be characterized as ‘hybrids’, and essentially are descriptions of existing knowledge, rather than extrapolations. A consequence of this approach is that error margins and uncertainties are never specified, and on the whole the models lack a clear formalized methodology for including both ‘hard’ and ‘soft’ knowledge.
The current chapter discusses the possibilities of improving the applied methodology by focusing on the formalization of the inclusion of ‘expert judgment’ or subjective knowledge into the mapping. Two methodological innovations are suggested for this: the application of a multicriteria decision-making framework to the modeling, and the use of Bayesian statistics for developing the knowledge base needed for the model. This methodological framework is tested on a case study in the municipality of Ede (figure 4.1), where the model of Soonius and Ankum (1990) was originally applied.

P. Verhagen (2009). Predictive models put to the test. In H. Kamermans, M. van Leusen & Ph. Verhagen (eds.): Archaeological prediction and risk assessment. Alternatives to current practice. Leiden University Press, Leiden (ASLU 17), pp. 71-122.

by Philip Verhagen

Available as chapter 7 of my thesis

In 2002, the research project ‘Strategic research into, and development of best practice for, predictive modelling on... more In 2002, the research project ‘Strategic research into, and development of best practice for, predictive modelling on behalf of Dutch cultural resource management’ (Kamermans et al., 2005; van Leusen et al., 2005) started out by identifying the research themes that were considered to be of great importance for the improvement of the quality of predictive modelling in the Netherlands . One of these themes concerned testing of currently used predictive models, as a means to assess their quality. Very little seemed to be known about the best way to test a predictive model, and in practice tests that have been carried out were limited, and have not been used in a systematic manner to improve the predictive models. At the same time, more and more data sets have become available for predictive model testing because of the enormous increase of archaeological research carried out in the Netherlands, following the ratification of the Valletta Convention. It was therefore decided that the subject should be studied in more detail in the second phase of the project. The current chapter is the result of this more detailed investigation, which has been carried out between January and July 2005.
The research questions defined for this study are:
- Can we identify testing methods that can measure predictive model quality in an unambiguous way, and that will allow us to say whether model A is doing better than model B?
- If these methods exist, what kind of predictive models do we need in order to apply them?
- Is testing really necessary? Can we perhaps deal with the issue of predictive model quality by defining statistical probabilities and confidence limits?
- And finally: do we have the data sets that will allow us to carry out predictive model testing in a rigorous way? And if not, how can we generate these in the future?

These questions are addressed through a review of existing testing methods for archaeological predictive models, that have appeared in and outside archaeological literature since the late 1980s. This analysis is followed by an exploration of the potential of currently available archaeological data sets in the Netherlands for predictive model testing purposes. In teh next section , the question of suitable models will be addressed: what testing methods are applicable to different types of models, and can we identify the model types best suited for testing? In the last section, the conclusions and recommendations of the study will be presented.

A Petrographic Approach to Sand-tempered Pottery Provenance Studies: Examples from Two Hohokam Local Systems

by James Heidke

Co-authored with Elizabeth J. Miksa and Henry D. Wallace

Some Statistical Remarks on Classification of Weapons Deposited in Graves

by Dan Stefan

When they are discovered in tombs, weapons provide complex assessments on the deceased and the society from which they... more When they are discovered in tombs, weapons provide complex assessments on the deceased and the society from which they come. Largely speaking, the structure of funerary inventory is a result of a wide variety of phenomena (fashion, sex, status, beliefs, technology, etc.) or circumstances (natural hazards, disease, damage caused by humans). It is also recognized their role as bearers of social or ideological messages. A complete “reverse engineering” is most often impossible, however traditional analysis based on typology and the study of the distribution of types may reveal a part of the basic funerary behavior patterns valid for a community at a time. To refine the interpretation we would have to overcome barriers of the traditional definition of types, seen as groups of interrelated objects with consistent similar features which may have social and/or spatial and temporal significance.
The present paper is devoted to a vision of organizing knowledge from the funerary space as a process by which objects associated with the deceased are differentiated in natural categories (rather than subjective) so that specific relationships can be subsequently recognized. This cognitive process will be illustrated based on the analyses of the weapons deposited in some Iron Age graves from the sub-Carpathian area.

Rezidualita v městských souvrstvích - Residuality in Urban Stratifications

by Karel Nováček

in: Šmejda, L. - Vařeka, P. (Eds.): Sedmdesát neustupných let. 1. vyd. Plzeň: Aleš Čeněk s.r.o., 2003.

One can suppose that the residuality is a function of the land use intensity. Migration and deposition of artefacts... more One can suppose that the residuality is a function of the land use intensity. Migration and deposition of artefacts (and ecofacts) into considerably younger deposits is by no means rare, random phenomenon in the urban archaeological environment- The study of this complex process cannot be restricted to the concrete, individual, chronologically distinctive finds which might be recognized with the naked eye. Identification of differences among the archaeological assemblages in spatio – temporal perspective by quantitative methods is decisive for study of the residuality: heterogeneity and fragmentarisation of assemblages seems to be the most important indicator. Besides the measure of diversity and seriation used in our case–study a further analytical potential of Harris stratigraphic diagramms has been also verified. It is to be stressed that the study of residuality and infiltration in complicated urban stratifications makes some demands on the method of excavation: it would need to take heed of the measure of volume of deposits and to modificate the description of deposits so as to obtain more exact data for their genetic clasification and quantitative analysis.

P. Verhagen, M. van Leusen, B. Ducke, A. Millard & H. Kamermans, 2011. The Bumpy Road to Incorporating Uncertainty in Predictive Modelling. In: Jerem, E., Redö, F. & Szeverényi, V. (eds.): On the Road to Reconstructing the Past. Proceedings of the 36th Annual Conference on Computer Applications and Quantitative Methods in Archaeology, CAA 2008. Budapest, April 2-6, pp. 569-576.

by Philip Verhagen

One of the key problems of predictive modelling is the lack of tools to incorporate and map the uncertainties of the... more One of the key problems of predictive modelling is the lack of tools to incorporate and map the uncertainties of the predictions made. Without explicit description of the varying quality of the archaeological and environmental data, statistical methods risk making inaccurate predictions. Hence, lacking adequate descriptions of bias and error, predictions often rely on expert judgement. But can expert judgement be quantified in such a way, that predictions can be made that will respect the experts’ views, and at the same time refl ect the uncertainties in the experts’ opinions as well as in the available data? This paper reports an investigation into whether expert views can be quantified and incorporated in statistical predictions, for which we tested two potentially useful techniques, Bayesian inference and Dempster-Shafer theory.

Qualitative landscape theories and archaeological predictive modelling – a journey through no man’s land?

by Dorothy Graves McEwan

in review; Journal of Archaeological Method and Theory.

A new method using qualitative (experiential) landscape theory to test and reinterpret GIS-based, quantitative... more A new method using qualitative (experiential) landscape theory to test and reinterpret GIS-based, quantitative archaeological predictive models is proposed. A case study is used to demonstrate the method in practice. Modifications of models are suggested after considering qualitative and quantitative investigations at the case study. The result is the transformation of statistical data about the distributions of archaeological sites into archaeologically meaningful interpretations of spatial data.

The use of predictive modelling to target Neolithic settlement and occupation activity in mainland Scotland

by Dorothy Graves McEwan

Journal of Archaeological Science, Volume 38, Issue 3, March 2011, Pages 633-656.

A central aim of Neolithic studies in Europe is to understand settlement or domestic activity of the first... more A central aim of Neolithic studies in Europe is to understand settlement or domestic activity of the first agriculturalists. In Scotland, the continued absence of unambiguous Neolithic settlement or domestic occupation activities on the mainland remains an unresolved issue. It is proposed that areas attractive for settlement and occupation activities may be identified by constructing GIS-based models of sites believed to be related to settlement or occupation activities, including chambered cairns, timber halls, and sites containing pit-digging episodes. Statistical analyses undertaken in this process suggest new insights into the locations of the chambered cairns, timber halls, and pits. Finalised models are constructed and desktop assessments are conducted to examine model performance. Research priorities for improving significant environmental variables driving the current models are identified.

Džbány českého eneolitu - Jugs in the Eneolithic of Bohemia

by Petr Krištuf

Krištuf, P. 2005: Džbány českého eneolitu. In: E. Neustupný - J. John (eds.), Přispěvky k archeologii 2, Plzeň: KAR FF ZČU, 69-118.

This paper deals with the structure of Eneolithic society in the territory of Central Europe. I studied these problems... more This paper deals with the structure of Eneolithic society in the territory of Central Europe. I studied these problems in terms of jugs from Bohemia. I infer that Eneolithic jugs were made for drinking of alcoholic beverages during the so-called “drinking feast.” Such drinking events structured prehistoric societies.
I assume that social differentiation of the Eneolithic society can also be studied in terms of decoration and morphology of jugs. The differentiation in social status between man and women is assumed for the Bohemian Eneolithic period. This differentiation is apparent, for example, in the burial rites of the Corded Ware Culture or Globular Amphora Culture. My aim is to demonstrate that this differentiation is also reflected in the size and decorative patterns of jugs.
I apply the principal component analysis to find structures in the collection of Eneolithic jugs. This method, working with linear vector space, finds structures that could not be discovered by the “traditional” approach.
One of the main results of my investigation is that volume of male and femalejugs differs in some Eneolithic cultures. I have also detected other regularities some of which seem to be chronologically determined.