***** To join INSNA, visit http://www.insna.org *****
It is not only sociologists that find themselves perplexed by the review
process at Nature and Science. Last year there was an article in Nature
that will be useful to any of you tasked with teaching statistics,
because it is a beautiful example of how NOT to use Regression"
"Momentous sprint at the 2156 Olympics? Women sprinters are closing the
gap on men and may one day overtake them." NATURE|VOL 431 | 30 SEPTEMBER
My husband, a muscle physiologists was outraged by the lack of any
intuition about physiology; I by the statistical violations, so we wrote
a reply: it was triaged. The paper extrapolated Olympic 100 meter times
for men and women to show that by the year 2156 woman will outpace men.
The linear model was not the best fit, but it was the "easiest and most
parsimonious." (We reevaluated the data using a spline which was the
best fit.) The researchers seemed to attribute the faster finishing
times to changes in physiology that occurred this century rather than
improvements in selection, nutrition, technology or training.
The authors were not physicists.
Below is the raw text of the reply (figures and citations not included).
If anyone is interested I can email you the PDF file.
Gender gap revisited: a physiological glass ceiling?
Jon-Philippe K. Hyatt and Elisa Jayne Bienenstock
The prospect that women will outpace men in future Olympic 100-meter
contests, as recently reported in Nature1, would indeed be a momentous
achievement in overcoming competitive, sociological, and physiological
barriers. Today, it is not uncommon for women to stand atop the podium
in ultra-endurance events: in 2002 and 2003, Pam Reed ran to victory
over all entrants in the 218-kilometre Badwater Ultramarathon (Death
Valley, USA), suggesting that the gender gap in competition is narrowing
rapidly2. There is no argument that 100-metre sprint times for men and
women also have converged; however, the predictions put forward by Tatem
et al.1 are untenable from a historical, statistical, and physiological
Assessment of gender-specific performance as a linear continuum (winning
time x year) assumes consistent historical variables. Since the early
1980s, improvements in clothing, shoe technology, track surfaces, and
event-specific training have created advantages for today's Olympians.
Additionally, society's perception of women in sports advanced just
prior to this period as evidenced, for example, by the adoption of Title
IX in the United States in 1972. Assessing performance for each gender
must control for these recent technological and social advancements.
This is evident in our evaluation of the Olympic data: when women's
Olympic times (ref 1, see supplementary information)
are re-examined, curve- fitting models show that the year 1985
delineates past and present Olympians (Fig. 1). Our new best- fit curve,
e.g. a spline, actually shows women's time increasing after 1985, which
is clearly an artifact resulting from the limited number of 100-metre
times from 1985-2004.
Using one winning Olympic time per four years as the primary basis of
comparison limits our view of the potential held by each gender
especially since the Olympics do not necessarily represent the best
100-metre times for men and women in the last 20 years. Expanding the
analysis using the 100-metre times from the World Championship final's
heat (www.iaaf.org/WCH03/history/index.html; for data set, see
supplementary information) from 1987-2003 shows that performance for men
and women have improved marginally, if at all (Fig. 1). Statistical
analyses reveal that, in fact, there is a poor relationship between 100-
metre times and year, supporting the notion that peak performance for
this event has reached a plateau for men and women. That sprint
performance and time (year) are unrelated in the past two decades
indicates that it is erroneous to predict future human success based on
past and present winning times3.
The argument put forth by Tatem and colleagues is also physiologically
incorrect, as it ignores limitations that exist for the human body. For
instance, the logic of their linear continuum suggests that, given
enough time, men and women will outrun quadrupeds, automobiles,
airplanes, and even light, eventually crossing the finish line in less
then zero seconds. Will women ever out-sprint men in the 100-metre
event? Our examination of 100-metre times from recent years suggests
that we are reaching our physiological limits of performance2,4,
although there may be innovations through science and technology that
will allow us to embrace such a fantastic achievement when it happens.
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.