Biology Reference
In-Depth Information
osteoporotic senior adults. Attention to the integrity of the bone surface (e.g., flaked, cracked,
or peeled cortical bone), bone completeness (e.g., missing distal or proximal ends of long
bones, missing ribs), or bone element absence (e.g., no phalanges) will alert the scholar to
question the representativeness of the age-at-death profile as well as to the possibility of
pseudopathologies. For example, an eroded outer cortex that exposes the normal internal
cancellous bone may easily be mistaken for a pathological condition. Awareness of preserva-
tion problems, use of a hand lens, and a thorough photographic record are key factors in the
gathering of objective and accurate biocultural information.
Sample Size and Statistical Analysis
The size of the skeletal sample will determine the method of bioarchaeological inquiry. If
the objective is to compare the prevalence of pathological conditions between samples,
between the sexes, between statuses, or between subsistence for significant difference, then
the samples should be large enough to enable simple statistical tests. See DiGangi and Moore
(Chapter 2), this volume, for a discussion on sample size. Two commonly, but not exclusively,
used statistical tests are chi-square and Fisher's exact test. Conventional wisdom indicates
that the former is used when the expected number in any cell of a contingency table (for
example, two rows by two columns) is greater than five. Otherwise, and preferentially,
Fisher's exact test is used. Both statistics test the hypothesis of no difference (i.e., a null
hypothesis) between the samples' prevalence. Both statistics generate a calculated value
(p) which, if less than 0.05 (i.e., a confidence interval of 95%) rejects the null hypothesis.
15
Besides statistical software packages that can be purchased or are available through educa-
tional institutions, there are a number of free statistical programs on the Internet that calcu-
late a wide variety of tests. There are many ways to pose health status questions and, with
advice from statistics experts, information can be segregated and tested to answer specific
questions (exemplified below).
If the sample consists of less than (say) a dozen individuals, the analysis is primarily
descriptive. This does not mean that the individual is simply inventoried and assessed for
the presence of pathologies as well as other particular information (e.g., discrete traits and
metric data). Assessment requires more contextual information (osteobiography) about the
individual such as mode of interment, grave accompaniments, and burial location within
the bonded cemetery area and within the community.
This information, in turn, is compared to and contrasted with in two ways. First, if avail-
able, the osteobiographic information is compared to prevalence patterns generated from
bioarchaeological (“population”-based) assessments of larger skeletal samples drawn from
the same archaeological context (e.g., another local site from the same time period or cultural
phase) as the small sample. This would help ascertain how typical the individuals are relative
to the health patterns of the comparative sample. For example, if the study is a single case of
a high-status individual who happens to be the only case of rickets in the entire cultural
phase, a closer examination of the individual's social role is merited. The second way a small
15
The null hypothesis is the default condition or expectation of any given test. It can be either supported or
not supported by the outcome of a statistical test. Refer to DiGangi and Moore (Chapter 2), this volume.
