Biology Reference
In-Depth Information
FIGURE 7.5
Maps of (a) Sudanese Nubia, Sudan, (b) M´ rrope in the Lambayeque region of Peru, and (c) west-
central Tennessee, United States.
As cribra orbitalia is a childhood condition (
Stuart-Macadam, 1985
; Stuart-Macadam and
Kent, 1992), the authors examined the prevalence across all the subadults. There was a steep
increase in frequency across the infant (0
e
1 year, 7/41 or 17.07%) and weanling (2
e
3 years,
17/34 or 50%) categories with prevalence peaking (75
e
78%) across the next three categories
(4
e
6 years, 7
e
9 years, and 10
e
12 years) and dropping to 50% in the 13
e
15-year-old category.
The prevalence across age categories can be tested for significance using a simple 2
2
contingency table (e.g., Fisher's exact test). For example, the proportion of weanlings with
cribra (17 with, 17 without) relative to the 4
e
6-year-old age category (42 with, 15 without)
is statistically significant (p
0.0257]) (see above).
Because of small sample sizes, sex differences could only be addressed in the collective
sample, not by temporal unit. Despite the sample being pooled, sample size segregated by
sex across five adult age categories dramatically reduced the number of individuals in
each age group. For example, there were only 6 males and 11 females in the 16
e
20 adult
age category and 11 males and 14 females in the 21
e
30 adult age category. The interpretive
question would be: Are 3/6 cribra (50%) cases in the young males compared to 3/9 (33%)
cases in females spurious results or culturally significant? Rather than speculating about
the meaning of what in all likelihood could be mathematical sampling error, Mittler and
Van Gerven, exemplifying the analytical effectiveness of using multiple stress indicators,
marshaled previous research on the Kulubnarti samples on subadult long bone growth
(Moore et al., 1986) and linear enamel hypoplasia data (
Van Gerven et al., 1990
) to argue
that young males are demonstrably stressed, therefore the results in this study are probably
not spurious. Had they pooled the young adult sample to a 16
e
30 years age-at-death cate-
gory, the sex differences (6/17 male cases, 7/23 female cases) would not have been statisti-
cally significant (p
0.05 [p
<
¼
1.0000) and the authors would not have probed further. This
¼
