Biology Reference
In-Depth Information
a lower percentage of the correct identification of sex from the Phenice characteristics, but
this study only examined a small sample of 50 individuals and included multiple observers,
some of whom were novices. If we have at least some “known” sex skeletons from a popula-
tion based on either DNA or Phenice characteristics, then we can use these as a calibration or
training sample (in other words, as the reference sample) in order to sex additional skeletons
(from that population) for which DNA or the Phenice characteristics are unavailable. To illus-
trate how this can be done we use an example from
Konigsberg and Hens (1998)
.
Estimating the Sex Ratio
To estimate the sex ratio within a skeletal sample, we can simply estimate the proportion of
one sex (either male or female) within that sample.
Konigsberg and Hens (1998)
gave a tabu-
lation of how 114 crania from the Averbuch Site (Mississippian period, Middle Tennessee)
would have been sexed on the basis of brow ridge morphology (see
Table 11.1
). Sixty of these
114 crania had previously been sexed as being from males on the basis of the Phenice char-
acteristics from the associated pubic bones, while the remaining 54 individuals had been
sexed as females on the Phenice characteristics. Note that the sexing from the brow ridges
was done without knowledge of the sex from the Phenice characteristics. Further, Konigsberg
and Hens originally selected the 114 individuals so that the identification of sex from their
pubic bones was unambiguous. Consequently, these 114 individuals can be treated as
a known sex reference sample. From the 60 males Konigsberg and Hens found that on the
basis of the brow ridges they would have sexed 51 as being male, 5 as female, and 4 as inde-
terminate. From the 54 females they would have sexed 32 as being female, 16 as being male,
and 6 as indeterminate. Now imagine that we score the brows in a sample of 100 crania from
the same population (but without the ability to observe the Phenice characteristics) as 35
“male,” 11 “indeterminate,” and 54 “female.” The immediate demographic task is not to
attempt to determine the sex for each cranium, but instead to estimate the proportion of
the sample that is either male or female. Following on from this, we can return to the problem
of estimating sex for each cranium. For this example, we will estimate the proportion of indi-
viduals that are male, which we will show as p
m
. The proportion of individuals that are
female is then p
f
¼
1
e
p
m
.
To estimate p
m
we will use the method of maximum likelihood estimation (MLE). There
is a relatively gentle introduction to MLE by Purcell
3
and
Myung (2003)
gives a nice tutorial
using MATLAB. MATLAB is a proprietary program, so in its place we use “R,” which has an
TABLE 11.1 Tabulation of Estimated Sex from the Brow Ridge against the
Phenice (1969)
Characteristics of
Sex from the Pelvis. 114 Individuals from the Averbuch Site (
Konigsberg and Hens, 1998
)
Indication from Brow
Phenice
Characteristics
“M”
“?”
“F”
Male
51
4
5
Female
16
6
32
