Biology Reference
In-Depth Information
Trotter and Gleser (1951a)
reevaluated Rollet's much earlier data and found similar average
stature loss of about 1.2 cmover 20 years. Thismethoddoeswell until age 70 and then it begins to
overestimate stature (
Galloway, 1988
). On a study of 8000 military personnel,
Giles and Hutch-
inson (1991)
recognized an age-related bias, inwhich stature begins to decrease between 40 and
44 years. They calculated thatmales lose on average 1mm/year and females lose 1.25mm/year.
After the age of 75, the age decrease accelerates to 1.4 mm/year for males and 2.0 mm/year for
females (
Giles andHutchinson, 1991
). Gallowaydetermined the age of stature reductiononset at
45 after surveying 550 individuals of older ages for maximum reported height when they were
age 25 andmeasuring their current stature. There was a progressive loss in stature in both sexes,
though her results had a relatively weak correlation.
Galloway (1988)
recommends using
a revised correction factor for age to account for age-related stature decrease, as opposed to
the one by
Trotter and Gleser, (1951a)
. See both equations in
Box 6.6
.
Age-related decline in stature typically reflects compression of soft tissues and
not the length
of the bones
(though osteoporotic fractures of the vertebrae can increase stature loss) (
Galloway,
1988
). Individuals are unlikely to change the initial stature on their driver's licenses,
regardless of height increases or decreases (
Galloway, 1988; Giles and Hutchinson, 1991;
Willey and Falsetti, 1991
). Thus,
Ousley (1995)
finds that FSTAT, as self-reported maximum
stature on ID cards as a young adult, may more accurately reflect the calculated stature
from the long bones, because the long bones do not change in length as a result of age. FSTAT
is also more readily available than MSTAT.
16
To be conservative for forensic cases,
Galloway
(1988)
recommends providing stature estimation both with and without age corrections.
Wilson and colleagues (2010) conducted an evaluation of stature estimation using the
existing data in the Forensic Databank (FDB). When
Ousley (1995)
had used the same data
15 years earlier, there were fewer individuals in the FDB. With a larger sample size, Wilson
and colleagues (2010) were able to test the accuracy of FSTATand found that their ASTAT (or
Any Antemortem Stature Available: cadaver, measured, or forensic) performed equally well
as FSTAT. They used traditional inverse calibration and calculated prediction intervals and
confidence intervals to compare precision and accuracy. The mean squared error represents
differences between the actual/reported and the predicted stature, which was used to test the
predictability power of the equations (
Wilson et al., 2010
).
Statistical Methods: Another Case of Apples to Oranges
Although most studies have used least squares regression, Konigsberg and colleagues
(1998) point out that it is actually
inverse calibration
regression, because it solves for stature
as the dependent variable (
y
stature). When Pearson first introduced regression for stature
estimation, he also set the trend for using the inverse calibration method (
Pearson, 1899
). In
most other regression analyses in science, the independent variable
x
will have an effect on
the dependent variable
y
. When stature is used as the
y
variable, it implies that the bone
length measured
causes
stature, which is an improbable causal relationship. Depending on
the context, other statistical methods may actually be better (
Konigsberg et al., 1998
).
In allometry literature, stature
is explained by
the size of particular organs/bones (i.e. stature
ΒΌ
16
You have probably had your own stature measured many times and by many clinicians, but your driver's
license stature is the most easily accessed stature.
