Biology Reference
In-Depth Information
(not rate of change) plotted by age have limited explanatory power (
Lovejoy et al., 1990;
Saunders and Hoppa, 1993
).
Klaus and Tam (2009)
determined that after age five, the post-
contact children did indeed display a slower rate of growth that was statistically significantly
different.
A way of measuring population growth (fertility) is to comparatively assess the ratio of
surviving reproductive age women (30
years of age)
to subadults who survived
þ
infancy and weaning (5
years of age) (
Buikstra et al., 1986
). Females who are undernour-
ished and overworked have difficulties conceiving and carrying pregnancies to term (Ellison
and O'Rourke, 2000). In this study, small adult total sample sizes made a separate assessment
of female health problematic so Klaus and Tam utilized the fertility ratio D
30
þ
/D
5
þ
as a proxy
for the health status of adult females (2009). A ratio was generated for each time period and
a straightforward statistical test (z-ratio
19
) did indeed indicate a statistically significant drop
in fertility with Spanish contact.
Klaus and Tam had extensive problems with sample size within and between the temporal
groups necessitating broad age categories. Nevertheless, for LEH assessment in the 15
e
24.9
years age-at-death category (males and females) there were 55/69 pre-contact (79.7%)
individuals versus 4/9 (44%) post-contact individuals. The total sample size for assessing
periostitis prevalence for the pre- AND post-contact horizons in the 35
e
44.9 (males and
females) age-at-death category was small, only 11 individuals. How should analysis proceed?
The authors opted to use a method not affected by sample size that is infrequently adopted
bioarchaeologically but is often utilized in health-related studies (and horse racing): the odds
ratio (
Waldron, 1994; McHugh, 2009
). This statistic assesses the ratio of two outcomes (e.g.,
presence versus absence), not the raw frequency (percent of a particular outcome), between
two groups. The odds ratio (OR) statistic performs a comparison between two samples (e.g.,
pre- versus post-contact) of the quotient of the outcomes of two events or conditions (e.g.,
compare pre-contact presence
O
absence with post-contact presence
O
absence). The
authors utilized a statistical package (SAS 9.1) to test for a significant difference between
all of the paired comparisons.
20
The OR results were porotic hyperostosis and (particularly)
periostitis expectedly increased in prevalence in the post-contact period but LEH showed
a decline. The authors consider three plausible explanations for the LEH decline: sampling
bias, the shorter duration tooth development within which to document enamel-damaging
stress episodes, and (most likely) a historically documented epidemiological shift to acute
rather than chronic health problems.
þ
Case Study: Treponemal Disease and Sedentism
Most specific infections are recognized by no more than a few cases in any one skeletal
sample. However, if a multiple site sample in close geographic proximity and similar social
and economic contexts can be generated (a meta-analysis), a specific disease can become
a tool of bioarchaeological inquiry. One of these circumstances is the prevalence of trepo-
nemal disease in an eight-site sample (n
581) that straddles the Late Archaic (circa 3000
¼
19
This can be calculated at websites such as
http://faculty.vassar.edu/lowry/propdiff_ind.html
.
20
The odds ratio can be easily calculated using free software available online. For example:
http://www.
