Page 87 - IJPS-7-2

International Journal of
Population Studies Modeling archaeological mortuary assemblages

normal mortuary assemblages and asking the question of investigation (Wood et al., 1994; 2002). This relationship
whether the MTC and CI assemblages are clearly different is viewed through a specific filter introduced by the choice
from a horticulturalist model life table, thus ruling out of a model life table and specific demographic model by
the simplest possible alternative explanation for their which to analyze these data (Hill & Hurtado, 1996; Keckler,
accumulation. The focus of the paper is to test the null 1997; Wood et al., 1994).
hypothesis that these assemblages accumulated through An acceptable consideration of the relationship
such a standard mortality process. A rejection of this between what we observe in a mortuary assemblage
null hypothesis would securely establish the standing of and what we expect must, therefore, present a specific
these sites as examples of assemblages accumulated due to strategy for dealing with bias in the probability of
human sacrifice.
observing deaths and the utilization of a model life table
1.2. Background with high plausibility. In this paper, we address both
concerns in comparing the MTC and CI assemblages to
1.2.1. Mortuary assemblages
a reference mortality process associated with traditional,
Mortuary assemblages from archaeological sites should horticulturalist populations. First, an expert-based
not be assumed to reflect the typical J-curve age at death model of preservation bias was employed that explicitly
distributions that is expected in the mortuary assemblages addresses the probability of observation of each death in
of living populations (Chamberlain, 2006; Weiss, 1973). the assemblage. This model includes an examination of
Differences may exist due to unanticipated disasters such both the average anticipated probability of observation
as epidemics, warfare, and human sacrifice (Chamberlain, as well as high and low variants that capture a range of
2006) or they may accumulate for taphonomic reasons uncertainty. These estimates were combined in the context
related to culture-specific mortuary practices (Saunders of a normal distributed Monte Carlo model (Graham
et al., 1992; Scrimshaw, 1984), soil conditions (Gordon & Talay, 2013; LeMieux, 2009) of the probability of
& Buikstra, 1981; Haglund & Sorg, 1997; 2002) or even observations that is reported in this paper and, therefore,
as artifacts of excavation methods (Paine & Harpending could plausibly be re-adjusted by readers using our tables
1998). For all of these reasons, bioarchaeologists have if they prefer to explore different levels. The model grows
long been suspicious about the assumption that mortuary out of the work of Saunders et al. (1992; 2002) in which
assemblages directly represent mortality processes (Angel, comparisons of cemetery populations and parish records
1966; Bocquet-Appel & Massett, 1982; Weiss, 1973). were made that provided explicit starting estimates for
McCaa (2002) goes as far as to call this the “Whopper considering these issues. It is based in the theory behind
Assumption.” Horvitz-Thompson estimators (Horvitz & Thompson,
In addition to the challenges associated with differential 1952), in which observed data may be up-weighted to
preservation of materials, mortality analysis often makes reflect unobserved data in calculating summary statistics
very specific assumptions about demographic non- (Longford, 2005) Here, fellow colleagues experienced in
stationarity, differences in frailty across subsets of the bioarchaeology and paleodemography were consulted by
population, and the role of selective mortality (Wood the author to form estimates of the average, high, and low
et al., 1992) that is not always acknowledged in any ranges of bias that could be present and these estimates
given analysis in the literature of bioarchaeology and formed the basis of our analysis.
paleodemography (Keckler, 1997). A common approach To adequately choose a model life table for developing
has been to use the strategy of “model life-tables” expected distribution of age at death for each site,
(Bocquet-Appel & Massett, 1982; Paine & Harpending, this method built upon the analysis of traditional
1996) to correct for deficiencies in bioarchaeological data. horticulturalist mortality patterns presented by Gurven &
This approach is common in demographic approaches Kaplan (2007). Their model comprehensively reviews what
to population analysis with incomplete data (Coale & is known about mortality in traditional, horticulturalist
Demeny, 1966; Preston et al. 1994; Shryock et al., 1980); populations and estimates the mortality schedule using
however, the choice of an acceptable model can have the Siler competing hazards model (Siler, 1979; 1983),
important impacts on any analysis (Hill & Hurtado, 1996; which parameterizes the life table by estimating effects
Howell, 1976; Keckler, 1997). associated with known differences in early-life, adult,

What we observe in a mortuary assemblage is not and old-age mortality patterns. As such, the Siler model
mortality; rather, it reflects a proportional relationship provides a biologically plausible basis for a model life
between mortality and the probability that we observe table that speaks to species-specific mortality patterns
any specific death in the context of bioarchaeological experienced as part of the human life course (Hill &

Volume 7 Issue 2 (2021) 81 https://doi.org/10.36922/ijps.v7i2.300