Dear Sportsci reader,
An innovative and dynamic school (department) within a regional Australian
University seeks a leader to drive its teaching and research in
biomechanics. The successful applicant needs to possess a PhD in
biomechanics or related field, possess a strong commitment to innovative
teaching practices and student progress, demonstrated leadership in
research, and be a team player.

Senior Lecturer (Level C) in Biomechanics
School of Health and Human Performance
Faculty of Arts, Health and Sciences
CQU Rockhampton
Full Time, Tenurable
Closing date for applications: 17th November, 2004.

Salary:  Senior Lecturer (Level C) - $68,772 to $79,092 per annum

The School of Health and Human Performance at Central Queensland University
(http://www.cqu.edu.au/) offers undergraduate and postgraduate degrees
and/or specialisations in Human Movement Science and Occupational Health
and safety in both internal and external mode.

School research interests include physical activity and it's role in health
promotion, clinical biomechanics, ergonomics, aging athletes,
thermoregulation in sport and the workplace, sport and exercise psychology,
sports performance, and the industrial athlete. The School is committed to
on-line teaching and learning, has a strong link to relevant regional
industry and is a leader in graduate placement. The School has a small
number of highly committed staff, possesses high staff and postgraduate
morale, and is dynamic.

Applications are called for a Senior Lecturer in Biomechanics at the
Rockhampton campus of Central Queensland University, commencing 1 February
2005. The successful applicant will be hard working, ambitious, and able to
work well both individually and in a team environment.

Applicants must complete an "Application Cover Sheet", refer to the
Position Description and address the Selection Criteria which can be
obtained at: http://dss-portal.cqu.edu.au/plato/vacancies.html or leave
your contact details on the 24-hour automated telephone line: (07) 4930
9974.  Applications should quote the vacancy reference number (2004/96) and
be forwarded electonically to: electronic-applications@... OR five
(5) copies posted to: Appointments Manager, Central Queensland University,
Rockhampton, Q'ld 4702, Australia by closing date.

Further enquiries regarding the position description and selection criteria
should be directed to Peter Reaburn PhD, Head of School of Health and Human
Performance, Central Queensland University, Rockhampton, Q. 4702,
Australia, Ph: 07 4930 6748, Fax: 07 4930 6875 or email p.reaburn@...

Peter Reaburn PhD
Head of School
School of Health and Human Performance
Central Queensland University
Rockhampton Q. 4702

Ph: (07) 4930 9813
Fax: (07) 4930 6875
E-mail: p.reaburn@...
http://www.ahs.cqu.edu.au/hhp/index.htm

"A University should be a tributary to the real world, not a sanctuary from it"

I had the wrong link in my previous message.  Alan actually sent me the
handbook, not the link.  I got the link off Google and put it into his
message, but it was the wrong one.  The correct link is
http://www.cochrane.org/resources/handbook/index.htm

Will

On October 21 I sent a request for help with meta-analysis of randomized
controlled trials.  Specifically, I wondered whether using only the
post-test means and SDs in the experimental and control groups was an
appropriate way to estimate the effect and its standard error (and hence
its weighting factor in the meta-analysis), and I wondered whether this method
of analysis was being used to get around the problem of missing p values
(especially "p>0.05", but also possibly "p<0.05").

Ed Winter didn't have an answer but made some useful comments.  Alan
Batterham pretty-much resolved the issue by finding the Cochrane Handbook
at the Cochrane site and sending me the relevant section.  Ian Shrier also
helped by sending my query to a Cochrane statistician.  His reply was
useful, but there were a few misunderstandings.  George Kelley was going to
help but I forstalled him by sending him Alan's reply, which he agreed
with.  The relevant responses are below.

In summary, yes, you can use only the post-test data for the meta-analysis
instead of the pre and post.  Doing so will lead to loss of precision when
the within-subject SD is less than the pure between-subject SD (the usual
situation for most of the measures we deal with in exercise and sport
science, in my experience) and better precision when the within is greater
than the between (the situation for some physiological measures).  There
are some other issues relating to randomization and matched-pairs analysis,
which those who are interested can find below.

Will


From: "Winter, Edward M" <E.M.Winter@...>

This is not much help for your specific query but your request seems to
lend support for the quoting of actual p values not simply > or < one's
chosen significance value - which is usually less than or equal to 0.05.

In my role as a reviewer of manuscripts, I usually find myself asking
authors to do just this.  It helps for instance to distinguish between p's
of 0.049 or 0.051 and lets the reader judge the practical significance of
outcomes - although calculations of effect sizes, confidence intervals or
the like would be preferable.  That of course takes us into debates about
hypothesis testing about which I think you and I are well aware.


From: Alan Batterham <spsamb@...>

I believe they address this issue, in part, in the Cochrane Handbook
http://www.cochrane.org/archives/handbook.doc . Excerpt below on change
scores vs final outcome scores.

In some circumstances an analysis based on changes from baseline will be
more efficient and powerful than comparison of final values as it removes a
component of between person variability from the analysis. However,
calculation of a change score requires measurement of the outcome twice and
in practice may be less efficient for outcomes which are unstable or
difficult to measure precisely, where the measurement error may be larger
than true between person baseline variability…

In practice a reviewer is likely to discover that the trials included in a
review may include a mixture of change from baseline and final value
scores. However, mixing of outcomes is not a problem when it comes to
meta-analysis. There is no statistical reason why trials with change from
baseline outcomes should not be combined in a meta-analysis with trials
with final measurement outcomes...  If the use of change scores does
increase precision, the studies presenting change scores will appropriately
be given higher weights in the analysis than they would have received if
final values had been used, as they will have smaller standard deviations.


My edited reply to Alan:

...Your message has prompted me to work out exactly what the standard error
is in the two situations.  I've written the following for eventual posting
to the list and/or use elsewhere.

Let sd be the within-subject error, and SD be the true between-subject
SD.  Let n be the number of subjects in the exptal group and n also in the
control group.  Assume no individual responses to the treatment.

Using difference of change scores, the standard error is root(4sd^2/n).
Using post-test scores, the standard error is root(2(SD^2+sd^2)/n).

So the break-even point occurs when sd=SD.  This point corresponds to a
reliability correlation of SD^2/(SD^2+sd^2) or 0.50.

As an important aside, if you use just the post-test data, the total sample
size (n+n) needs to be nearly 800 with the traditional approach to
sample-size estimation, for a smallest Cohen effect of 0.20.  Many
physiological measures have sd similar to SD, not because we measure them
poorly, but simply because we are all made of the same stuff at a
physiological level, and the variation we see in physiological measures
between subjects is due partly or largely to real within-subject
variation.  Those physiological variables might also be controlling
variables rather than controlled variables in homeostasis, so they are
naturally variable.  It follows that studies using ~10-20 subjects for such
variables are grossly underpowered, whether you analyze them for difference
in the changes or differences in the post-test.  You would need to
meta-analyze dozens of such studies to get acceptable precision for trivial
or small effects.

By the way, the sample size of 800 refers to 0.20 of the observed
between-subject standard deviation; that is, 0.20 of root(SD^2+sd^2).  If
you make it 0.2SD, as some would argue, the sample size needs to be even
larger.  Four times as large if sd=SD.  In the case of physiological
variables where sd is "real", I think it's sensible to use 0.20 of the
observed standard deviation.  After all, if we were all truly clones, SD
would approach zero and all Cohen effect sizes defined using only SD would
approach infinity.

...A further thought occurs to me.  For those variables where the
within-subject variation is greater than the true between, at first sight
there would appear to be little point in including any pre-test when you do
a controlled trial.  Why inflate your uncertainty?  But we do like to
randomize in a balanced fashion, in case there is an interaction between
the pretest score and the effect of the treatment.  Then I remembered some
simulations I did a few years ago, as a result of interacting with Greg
Atkinson...  If you randomize to the two groups in the usual pair-wise
fashion on the basis of the pre-test score of the dependent variable, you
can perform a "matched-pairs" analysis that counteracts to some extent the
additional error that arises from the usual repeated-measures analysis.  In
the simulations, I performed the matched pairs analysis in two ways, but
the one that matters is a paired t test between the groups for the
difference in the post-pre change.  I just checked the outcomes of the
analyses, and I found that the matched-pairs analysis pretty-much
completely offsets the loss of precision you get from the usual
repeated-measures analysis of post-pre changes.  There are only some subtle
differences that appear to be due to the degrees of freedom for the t
statistic.  In other words, for a reasonable sample size of around 20 or
more, you can use the pretest value in your analysis and not get any
inflation of the confidence interval, provided you use the matched-pairs
approach to the analysis.  This approach works only when SD<sd, as defined
in that previous email.  Matched-pairs analysis produces wider confidence
intervals than standard repeated-measures when SD>sd.

When you report your finding from a matched-pairs analysis, you would still
have to show the exact p value for this approach to feed through into the
meta-analysis.  No-one does matched-pairs analysis, and few report the
exact p value, so the whole thing is academic anyway...


Edited reply from Cochrane statistician, and my edited responses:

  >Regarding your friend's query I can offer a few thoughts:
  >
  >1. To me, the query confusingly mixes two different issues - missing
  >data and use of change-scores vs post-intervention measures. These two
  >issues are related, but not as intimately as the email suggests.

I was so surprised by the notion of all that info in pretests being thrown
away in the meta-analysis, that I presumed there had to be a good reason,
and missing p values seemed to me to be the only good reason.  I was partly
aware of the issue of better precision when the error of measurement is
large, which is why I stated "in general the sampling error will be much
larger than the error derived from the p value for the comparison of the
changes in the experimental and control groups".  I was right with "in
general", but I might have erred a little with "much" larger.  It depends
on the variable.  For some physiological variables most of the
between-subject SD is due to the within-subject SD, and for these just
using the post-test scores would be better.  But why measure them at all in
a pre-test?

  >2. Phrasing the problem in terms of missing P values strikes me as
  >misguided. First, P values should not be considered the main summary of
  >a statistical analysis.

Of course not, but it's often all that many authors give.  Please note that
I do not approve of p values.  In fact, I refuse to include them in any
publication with my name on it.

  >Second, for obtaining data for meta-analysis
  >such as those considered here (i.e. of continuous data), P values are
  >used as a last resort only when standard deviations cannot be obtained
  >directly.

I think you mean standard error of the effect statistic, not standard
deviations.  The SE is an SD, of course.

  >In particular, when your friend says "But you need the p value (or
  >better, the confidence limits) to work out the sampling error for the
  >effect, which you convert to the weighting factor for the effect in the
  >meta-analysis", I'd have put it as follows: the weight is obtained from
  >the standard error (or the variance) of the effect estimate, which is
  >usually best obtained from summary statistics such as standard
  >deviations.

No, I think your emphasis on standard deviations is a little--dare I say
it--misguided. :)

  >Sometimes these are not available and standard errors can be
  >obtained from confidence intervals, T statistics or exact P values.

You should first try to get the standard error from confidence intervals, T
stats, or exact p values, before you use pre and/or post SDs.  But if the
error of measurement is so large that you get better precision from only
the post SDs, then you use those.

  >As a
  >last resort, in-exact P values (such as P<0.05 or P>0.05) can be used to
  >impute standard errors, although this is not necessarily better than
  >imputing using results from other analyses or other studies in the
  >meta-analysis.
  >
  >3. I've had a quick look at the methods section of the Boule paper in
  >JAMA, and must say I don't recognise your colleague's concerns. The
  >authors have chosen to focus on post-intervention differences in the
  >outcomes they look at. This is a solid and reliable approach to
  >meta-analysing results of randomized trials. They do not discuss missing
  >P values, but quite reasonably estimate outcome standard deviations
  >using baseline standard deviations.

That last sentence can't be right.  They used post-intervention SDs to get
the SE.  They used pre-intervention SDs to get the denominator for the
standardized (Cohen) effect size, if I remember correctly.  Or if they
didn't, they should have, because the post-intervention SD in the
intervention group could be inflated by individual responses.  I am at home
and don't have the paper in front of me, and I won't be in to work for
another three days because we have a local conference at a hotel.  There is
also the issue of whether the denominator in the Cohen should be the
OBSERVED SD or the TRUE SD (free of measurement error).  In my view, it
depends on whether the "error" is due to technical measurement error or
true day-to-day within-subject variation.  But that's another story.  Does
Cochrane go into this issue?

  >4. I'm guessing your friend's main problem is that the authors look at
  >post-intervention differencess, but that they might have looked at
  >differences in changes-from-baseline. The standard error of a difference
  >in change-scores will often be smaller than that of a difference in
  >post-intervention scores (with corresponding smaller P values and bigger
  >weights). This does not necessarily indicate the the use of changes is
  >to be preferred.

Yes it does.

  >One reason is that standard deviations of
  >changes-from-baseline are often missing, and they need to be imputed
  >(occasionally based on in-exact P values).

But that was precisely my point!

Accepting Applications for Masters Candidates in Exercise Science at
Appalachian State University.  The Graduate degree offers three areas of
specialization 1) Research, 2) Clinical/Cardiopulmonary Rehabilitation,
and 3) Strength and Conditioning.  A limited number of graduate research
and teaching assistantships are available for qualified students.
Graduate assistants receive a stipend of $7,000 and are renewable for
the second year. A limited number of out-of-state tuition waivers are
available for students with outstanding academic credentials.
Interested candidates should contact Dr. Charles Dumke and consult the
exercise science graduate programs website.



Application Procedures

All applicants for admission to the Graduate School must:
(1) submit a completed application with a $35.00 non-refundable,
non-deductible application processing fee,
(2) submit three letters of recommendation,
(3) submit official GRE scores, and
(4) submit official transcripts of all previous academic work.

The application packet is not sent to the department for approval until
all official documents are on file. It is the applicant's responsibility
to monitor the application process. More information and application
materials are available here.


All applications and inquiries should be addressed to:
Cratis D. Williams Graduate School
Appalachian State University
B.B. Dougherty Administration Building
P.O. Box  32068
Boone,  N.C.  28608-2068

Procedures For Applying For Graduate Assistantships
1. Complete the admission procedures for graduate school as described on
this page.

2. Complete and return the application form for either a teaching
assistantship or research assistantship. You may complete both
applications if you are interested in either opportunity.

3. Send a resume that provides further information about the experiences
and skills you possess. Mail all materials to:

Charles Dumke, PhD, Graduate Program Director
Department of Heath, Leisure, and Exercise Science
Appalachian State University
Boone, North Carolina 28608

Last reminder for the Biostats Group to submit for ACSM Thematic Poster
session. Please email me to let me know you have sent in the abstract, and
then I will let ACSM know exactly who should be presenting at our session.

Ian Shrier MD, PhD, Dip Sport Med (FACSM)
Past-president, Canadian Academy of Sport Medicine
check out: www.casm-acms.org

Centre for Clinical Epidemiology and Community Studies
SMBD-Jewish General Hospital
3755 Cote Ste-Catherine Rd
Montreal, Qc  H3T 1E2
Tel: 514-340-8222 ext 7563
Fax: 514-340-7564

Young Investigator Special Issue (YISI) 1 of the Journal of Sports Science and Medicine is now available for you to access the abstracts and the full articles in http and pdf formats. The JSSM family is proud to start this new service to the Sport Science community and to young researchers. The Young Investigator Special Issue followed a normal peer-review process, except that there were no straight rejections in the first phase of review. We advised the reviewers of the Young Investigator Special Issue to proceed with constructive advice and remarks for all manuscripts. This offered a great opportunity for the Young Investigators to revise the manuscript, while at the same time contributing to the learning process.

We are very happy that we achieved our goal in a very short time. The JSSM Young Investigator Special Issue became ready one month before the planned schedule. Many expert referees wrote us and offered their help. Many other referees accepted our revision request immediately despite their busy schedule. The support of the supervisors was also substantial. At this point, we would like to thank all contributors to this first issue of the Young Investigator Special of the JSSM.

In the future, we hope that we can attract even more participants to share their knowledge with us, to write up their own work and to increase their motivation and confidence.

Finally, we hope that the readers will accept this Young Investigator Issue as a scientific source for research.
Enjoy it!

Young Investigator Special Issue Editor
Mustafa Atalay, MD, MPH, PhD

YISI Guest Editors
Eric Eils, PhD
Jukka A. Salmi, MSc, MPH
 

JOURNAL OF SPORTS SCIENCE AND MEDICINE, NOVEMBER 2004 - VOLUME 3,

YOUNG INVESTIGATOR SPECIAL ISSUE 1

TABLE OF CONTENTS

EDITORIAL
Young Investýgator Specýal Issue
Mustafa Atalay, Jukka A. Salmi and Eric Eils
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/edi/yisi-edi-1text.php

RESEARCH ARTICLE
The physiological responses of chronic heart failure patients to maximal strength test and a balke incremental test.
Itamar Levinger, Roger Bronks, David V. Cody, Ian Linton and Allan Davie
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/1/yisi1-1.htm

RESEARCH ARTICLE
An evaluation of the rearfoot posture in individuals with patellofemoral pain syndrome.
Pazit Levinger and Wendy Gille
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/2/yisi1-2.htm

RESEARCH ARTICLE
Muscular oxidatýve capacity in ovariectomized rats discussion on the endurance performance of female athletes with sports-related-amenorrhea.
Takahiro Sasa, Koichi Sairyo, Naoyuki Yoshida, Makoto Ishikawa, Mari Fukunaga and Natsuo Yasui
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/3/yisi1-3.htm

RESEARCH ARTICLE
Perceived exertion as an exercise intensity indicator in chronic heart failure patients on beta-blockers.
Itamar Levinger, Roger Bronks, David V. Cody, Ian Linton and Allan Davie
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/4/yisi1-4.htm

RESEARCH ARTICLE
Training alterations in elite cyclists may cause transient changes in glomerular filtration rate.
Chad D. Touchberry, Mark Ernsting, Greg Haff and J. Lon Kilgre
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/5/yisi1-5.htm
 
RESEARCH ARTICLE
Evaluating the influence of massage on leg strength, swelling, and pain following a half-marathon.
Lance G. Dawson, L.G, Kimberley A. Dawson and Peter M. Tiidus
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/6/yisi1-6.htm

RESEARCH ARTICLE
Variation in football players' sprint test performance across different ages and levels of competition.
Catarina Abrantes, Vitor Maçãs and Jaime Sampaio
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/7/yisi1-7.htm

RESEARCH ARTICLE
Exercise-induced hypervolemia may not be consequential to dehydration during exercise.
Bartholomew Kay, Brendan J. O'Brien and Nicholas D. Gill
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/8/yisi1-8.htm
 
RESEARCH ARTICLE
Understanding the role of behavior and cognitions in a group exercise setting.
Tina L. Shrigley and Kimberley A. Dawson
http://www20.uludag.edu.tr/%7Ehakan/sbtd/yisi/1/9/yisi1-9.htm
 

For more information on JSSM go to:
http://www.jssm.org or http://jssm.uludag.edu.tr

If you have any questions please contact me by e-mail at hakan@...
 
 
 

I received another message on the effects of lacto-ovo-vegetarian diet on body
building. This is from someone identified by another subscriber as '"the
authority" on nutrition for vegetarian athletes in the US.'

Chris Forbes-Ewan

THIRD MESSAGE ON LACTO-OVO-VEGETARIAN DIET AND BODY BUILDING

Much of my response will be my opinion in interpreting the literature,
or should I say lack of the literature.

First, It is my opinion that a healthy vegetarian diet should be able
to adequately support the training of all athletes. It is true that the
short term studies tend to suggest that men following a vegetarian diet
tend to have lower testesterone but this is not translated into lower
performance or strength.  I do not know of any source that can plot
testesterone levels against strength or performance and get the answer
that the higher your testosterone levels are the better.  The issue is
also complexed by the fact that testosterone is both free and bound,
that its effect is influenced by receptors, and that many other growth
factors (including insulin, insulin-like growth hormone, and growth
hormone) influence muscle metabolism and growth.

In my opinion, I feel much of the low testosterone may be from a
combination of a diet that is too low in fat and cholesterol and too
high in fiber. What is not known is by what mechanism this happens (as
you said) and whether the body adapts after so long on a vegetarian
diet.  There is a whole body of literature looking at low fat diets in
general and testosterone that might assist in the specific mechanism.

I am not sure what your diet is like so it is hard to go from here.  I
would strive to eat a diet that is a little higher in fat (30-35%)
rather than 15-20 if that is an issue and watch that your carbohydrates
are a balance of whole grains and natural more simple carbs. By this I
mean,  don't just eat whole-wheat pasta and fresh fruit but mix it up
with regular pasta, rice, juice and lower fiber fruits and veggies.

In response to Gavin's query about differences in standard deviations
(SDs)...  I'll comment on Gavin's specific issues first, then discuss the
more general issue of substantial differences in SDs.  I would love to hear
from others on this topic.

>Does anyone know if it is possible to find out if data from two groups are
>heterogenous just from reported means and SDs ?

Heterogeneity refers only to a substantial difference in the SDs, in my
understanding.  If distributions are non-normal, heterogeneity could
perhaps also apply to skewness, kurtosis, and even higher-order "moments",
but let's not go there.

>I have a number of studies
>I am reviewing which seem to have very heterogenous variances (double the SD
>for example)...

For anyone who just arrived from another star system, the variance is the
SD squared.  Statisticians work with variances, but what matters in the
real world is SDs.

>...and despite large between-group mean differences report no
>significant findings.

Small sample sizes, no doubt.  What are the chances that the true
difference between the means is clinically or practically worthwhile or
substantial?  That is the important question that an analysis via
statistical significance does not address properly.  I've built these
chances into my spreadsheets.

>Most authors give subject numbers, group means, S.D.s, S.E.Ms, t-stats or P -
>values and I was wondering if from these I can do an equivalent to a
>Levene's test for heterogeneity of variance?

The Levene "test" (I hate that concept) for heterogeneity of variance needs
only the SDs and the number of observations each came from (or more
exactly, the degrees of freedom for each variance).  It's not clear why you
want to test for differences in the SDs.  In my view you should always use
the unequal-variances t statistic, or its equivalent in mixed modeling,
when you estimate differences in means, regardless of what the test for
unequal variances shows.

Like tests for normality, tests for unequal variances are not particularly
useful, for three reasons.  First, what's the point anyway, if you are
going to use the unequal-variances t statistic?  Secondly, if the sample
sizes are small, there can be substantial differences in the SDs, yet the
test will not show significance, and most researchers will conclude
mistakenly that "there is no difference".  Thirdly, if the sample sizes are
large, there can be trivial differences in the SDs, yet the test can show
significance, and most researchers will mistakenly conclude "there is a
difference".  What matters is HOW BIG the difference is in the SDs, and how
uncertain you are about the difference, expressed as confidence limits for
the true difference and the chances that the true difference is clinically
or practically important.

Ah yes, but how big is BIG, when it comes to differences in SDs?  That's a
question that my colleagues and I have been asking for a few years.  Our
default at the moment is a factor of 1.15 (or a 15% difference).  That was
based on a consideration of sample size needed to quantify confidently a
difference or change in means.  The sample size is proportional to the
square of the appropriate SD, so a difference in the SD equivalent to a
factor of 1.15 would result in a difference in sample size of 1.32
(32%).  But that's pretty arbitrary, and anyway, sample size is not the
most important issue.  What matters is the impact on individuals, not
researchers.  So the rest of this message is about an attempt to come up
with a new way to decide on what's a substantial difference in an SD.

Let's start with the example I often use to explain magnitudes of
differences in means: IQs of men vs women.  Let's forget about the fact
that women on average have a higher IQ, and whether or not the difference
is substantial.  Instead, let's assume men and women have the same mean
IQ.  What about their SDs?  Apparently men have a larger SD than women.  A
larger SD implies that there will be more extremes in IQ amongst men than
women: the real geniuses and real dummies tend to be guys.  Let's make the
difference in SDs extreme.  Now draw any gal at random.  Then draw any
guy.  The guy will either be a genius relative to the gal, or a dummy.  So,
how big does the difference in SDs need to be before the guy is likely to
be substantially brighter or dumber than the gal, rather than very much
brighter or dumber?  That's the question.

Let's define substantially using my version of Cohen's scale:  0-0.2 is
trivial, 0.2-0.6 is small, 0.6-1.2 is moderate, and >1.2 is large.  Add the
negative values to this scale (-0.2 to -0.6 is small, etc.).  You end up
with seven size categories.  I've just devised a spreadsheet to show that
if you draw two people at random from the same population, the chances of
the difference between them falling into each category range from 11% to
20%, with all seven size categories adding up to 100%, of course.  Now draw
the second person from a population with a bigger SD.  The spreadsheet
shows that the chances of the difference between the two individuals
falling into each category change: chances of trivial, small and moderate
differences fall, chances of large differences rise.  When the second SD is
a factor of 1.11 bigger than the first, the chances of trivial and small
differences fall by a factor of 1.10, chances of moderate differences fall
by 1.07, and chances of large differences rise by 1.13.  Changes in the
chances or rate of something of this order (relative frequency or risk of
1.1) are arguably the smallest that matter or that are noticeable.  So it
looks like I have made a case for the smallest change in an SD being a
factor of 1.11, but let's make it 1.1 or 10%.  I've played around with the
spreadsheet in other ways and got pretty-much the same answer.

I may have gone down the wrong track somewhere here.  Or there may be a
better way to come up with the smallest difference in SDs.  Anyone else got
any ideas?

Will

Dear All,

Does anyone know if it is possible to find out if data from two groups are
heterogenous just from reported means and S.D.s ? I have a number of studies
I am reviewing which seem to have very heterogenous variances (double the SD
for example) and despite large between-group mean differences report no
significant findings.

Most authors give subject numbers, group means, S.D.s S.E.Ms, t-stats or P -
values and I was wondering if from these I can do and equivalent ot a
Levene's test for heterogeneity of variance?  Such stats are not commonly
(well, never) reported in the literature.

Any help prior to ACSM submission date would be greatly appreciated.


Gavin Sandercock
Research Centre for Health Studies
Buckinghamshire Chilterns University College
Chalfont Campus
Newlands Park
Gorelands Lane
Chalfont St. Giles
Buckinghamshire
UK
HP8 4AD
Tel: 01494 522141

Dear David, Chris and all on the list:

The comparison of men and women's world record, and their
progression is fascinating, which is why I guess the data (good, bad
or indifferent) often ends up in Nature, why it appears so
frequently in the media and why we all talk about it so much.

I think it's worth noting that when the prediction of world record
trends includes estimates of VO2max, lactate threshold velocity,
anaerobic power and capacity and VO2 kinetics (Peronnet and Thibault
(1989) J. Appl. Physiol., 67, 453-465), the predictions are
remarkably accurate.  These authors predicted running performance
from 100 m to the marathon, the results for males were as follows
(predicted and actual data for the year 2000):

Event: Predicted (actual) [%error]:
100 m: 9.74 s (9.79) [0.5]
200 m: 19.53 s (19.32) [1.1]
400 m: 43.44 s (43.18) [0.6]
800 m: 1:39.88 (1:41.11) [1.2]
1500 m: 3:25.45 (3:26.00) [0.3]
5000 m: 12:42.72 (12:39.36) [0.4]
10 000 m: 26:43.63 (26:22.77) [1.3]
Marathon: 2:05:23 (2:05:42) [0.2]

Of course, the limits of the physiological parameters in the model
are still unknown, and the authors assumed a near-linear improvement
in these parameters over time.  We could make educated guesses based
on data to estimate where the limits lie for both genders
(particularly for O2 transport and utilisation), but that rather
takes the fun out of the issue doesn't it?  I think a few quotes
best sum up my thoughts on these issues:

"My records will stand for a while.  The 400 m will be a tough
one to break ­ I wouldn't bet against it being broken in the next
decade…

The 200 m? Well, you could make a good bet that that person's
grandmother hasn't been born yet."
						 Michael Johnson

"The more restricted our society and work become, the more necessary
it will be to find some outlet for this craving for freedom. No one
can say, 'You must not run faster than this, or jump higher than
that.' The human spirit is indomitable."

Sir Roger Bannister

All the best
Mark


Mark Burnley PhD
Department of Sport and Exercise Science
Carwyn James Building
University of Wales, Aberystwyth
SY23 3FD
UK

This is what I sent to Chris yesterday.

David Bishop's commentary prompts my distribution to the group as a whole.

Professor Edward M Winter
The Centre for Sport and Exercise Science
Sheffield Hallam University
Collegiate Hall
Collegiate Crescent Campus
SHEFFIELD S10 2BP

Tel 0114  225  4333 (International +44  114  225  4333)
Fax 0114  225  4341 (International +44  114  225  4341)
e-mail e.m.winter@...
www.thecentreforsport.com

-----Original Message-----
From: Winter, Edward M
Sent: 18 October 2004 10:21
To: 'Forbes-Ewan, Chris'
Subject: RE: Male vs female running performance


Dear Chris

The manuscript by Tatem et al. was in the September 30 issue of Nature (Nature
431, page 525) with supplementary data and details
of the analyses on the Nature website.

The results suggested that women will run the same time for the 100 m in the
2156 Olympics.  This suggestion is based on a linear
extrapolation of data which gives rise to two immediate questions:  first, are
the data linear and second is extrapolation
warranted?

There is strong evidence that a linear model was appropriate although
exponential modelling would probably be biologically
preferable, but that extrapolation one and a half centuries hence might well not
be the wisest of forecasts.  For instance,
further extrapolation of a linear model would reveal when men and women would be
running 100 m in zero time . . .

I wasn't sure if this manuscript was tongue-in-cheek, even a hoax, but it has
caused debate.  It is similar to an earlier
manuscript from Nature by Whipp and Ward, quoted by Tatem et al., that women
would be running marathons in the same time as men by
about 2020.

Exponential modelling can perhaps be illustrated by the crossing-a-room
paradigm.  Start at one wall and facing the opposite one
and move half way across. Then move to half the remainder i.e. to three quarters
of the way across, followed by half the remainder
again to seven-eighths and so on.  You will always get closer to the opposite
wall but you will never actually reach it.

Time itself will tell whether or not the equality of performance forecasts are
accurate.  For my part each projection is unlikely
but that shouldn't prevent debate about the relative influence on performance of
biological and cultural factors.

A little over a week ago I sent a message to Sportscience about
lacto-ovo-vegetarian diet and body building. The original message is at the
bottom of this message.

I received three replies, one of which still requires follow-up on my part. The
other two are shown below.

FIRST REPLY:
Try a pubmed search for

Dorgan et al 1996; Berrino et al 2001

(Sorry I haven't got more details) - From memory both studies reveal a reduction
in Testosterone with diets that are low in saturated fats (less cholesterol -->
less Te?). A good thing for those with high risk of prostate cancer and perhaps
not such a good thing for some vegetarians.

[Note by Chris Forbes-Ewan]
I conducted the suggested search, and found two relevant articles:

(i) Dorgan JF, Judd JT, Longcope C, et al. (1996). Effects of dietary fat and
fiber on plasma and urine androgens and estrogens in men: a controlled feeding
study. Am J Clin Nutr. Dec;64(6):850-5.

(ii)  Berrino F, Bellati C, Secreto G, et al. (2001). Reducing bioavailable sex
hormones through a comprehensive change in diet: the diet and androgens (DIANA)
randomized trial. Cancer Epidemiol Biomarkers Prev. 10(1):25-33.

Dorgan et al. (1996) compared sex hormone levels in men eating either
high-fat/low-fiber or low-fat/high-fiber diets for ten weeks in a crossover
study. The high-fat/low-fiber diet was also a 'high saturated fat' diet (with a
P:S ratio of 1:3). They reported a tendency to increased plasma total
testosterone (13% higher) and a statistically significant increase for one
fraction of testosterone--the sex hormone binding globulin (SHBG)-bound
testosterone (15% higher) when the participants were consuming the
high-fat/low-fiber diet compared to low-fat/high-fiber. Dorgan et al. (1996)
also discussed evidence from previous studies raising the possibility that
vegetarians may have less bioavailable (non-SHBG) testosterone than omnivores.

Berrino et al. (2001) found that in postmenopausal women, a diet that is low in
animal fat and refined carbohydrates, but high in monounsaturated fats, n-3
polyunsaturated fats, GI and phytoestrogens, led to a reduction in serum
testosterone. They concluded that such dietary  modification "decreases the
bioavailability of serum sex hormones in hyperandrogenic postmenopausal women."



SECOND REPLY (RELEVANT EXTRACTS ONLY):
I get this question occasionally from endurance athletes, whose caloric
intake versus expense tends to visit negative values resulting in low
testosterone. Recently we tested 5 endurance cyclists and 1 female
runner. We found that their free salivary testosterone ranged from 34%
to 63% normal max reference range. While this represents not the whole
population, it may suggest that exercise expense has the potential to
tap into substrates that might have been en route to testosterone
production. I did a sweep through the literature and a hyperhealth
software program and found that there are some dietary and supplemental
protocols that may influence negative or positive testosterone values.


DIET AFFECTS TESTOSTERONE LEVELS

[1] HIGH INTAKE OF DIETARY PROTEINS increases levels of (free, unbound)
testosterone while LOW INTAKE OF DIETARY PROTEINS decreases
testosterone by increasing levels of sex hormone binding globulin (SHBG)
which binds to testosterone causing it to be unavailable to its
receptors; conversely high intake of dietary proteins lowers SHBG levels
which causes more free, unbound testosterone to become available for
binding to testosterone's receptors. Normally, 1%-3% of endogenous
testosterone is in its free, unbound state, while 44% is bound to sex
SHBG and 54% is bound to albumin and other endogenous proteins
testosterone that is albumin-bound IS AVAILABLE for tissue uptake;
approximately 56% of the body's testosterone is readily available for
uptake into tissues.) LOW PROTEIN DIET DOES NOT SUPPORT ANABOLIC
PRODUCTION OF TESTOSTERONE.

[2] REDUCED DIETARY FAT intake from 40%=>25% (dietary fats as a % of
total diet) results in an average 15% decrease in "free" testosterone.
Researchers determined that dietary fats should comprise 25%-30% of
total calorie intake in order to maintain optimal testosterone
production. LOW FAT DIET DOES NOT SUPPORT ANABOLIC PRODUCTION OF
TESTOSTERONE.

[3] SATURATED FATTY ACIDS increase endogenous testosterone levels.
Supplemental superunsaturated fatty acids (SUFAs) stimulate the Leydig
Cells of the testicles to synthesize testosterone.

[4] ISOFLAVONES such as DAIDZEIN (extract soybeans) increased in serum
testosterone (and beta-endorphin) in castrated rats. The same results
were obtained in chickens given total SOY ISOFLAVONES. VELVET BEANS are
reported to increase testosterone.

[5] OATS are claimed to stimulate the release of testosterone from its
"bound" state (with SHBG) to its "Free" state.

[6] GARLIC increases testosterone levels (due to the diallyl disulfide
content of garlic increasing luteinizing hormone (LH) levels.

SUPPLEMENTS AFFECT TESTOSTERONE LEVELS

[1] ACETYL-L-CARNITINE (ALC) increases plasma testosterone levels (via
its influence on acetylcholine neurotransmission in the striatal cortex
of the brain).

[2] MELATONIN prevents the age-related decline in testosterone
production. A 1995 study demonstrated that nightly supplementation of
melatonin to rats prevented age-related decline in testosterone
production. Melatonin supplemented rats had nearly 300% the level of
testosterone compared with rats that didn't receive melatonin.
Conversely, one study has demonstrated that supplemental HIGH
DOSE MELATONIN INHIBITS the production of testosterone - However, LOW
DOSAGE LEVELS OF MELATONIN (up to 3 mg per day) do not appear to
interfere with testosterone production. Males produce testosterone over
a 24-hour cycle - the highest levels occur in the morning. Testosterone
production and release does also occur qualitatively DURING SLEEP
meaning sleep deprivation leads to lower serum levels of testosterone.

[3] BETA-SITOSTEROL inhibit the conversion of testosterone to
dihydrotestosterone and estradiol (by inhibiting the 5-alpha reductase
and aromatase enzymes)

[4] BORON increases serum testosterone levels - boron does not raise
testosterone to higher than normal physiological values but does appear
to restore testosterone levels in older males to levels that they had in
their 20s and 30s.

[5] POTASSIUM helps to REGULATE plasma testosterone levels (potassium
deficiency can lead to sub-optimal plasma testosterone levels. Potassium
is high in whole plant foods such as fruits and vegetables.

[6] ZINC & SELENIUM are an essential cofactors for the endogenous
production of testosterone and supplemental zinc increases testosterone
levels that are depleted AS A RESULT OF ZINC DEFICIENCY. Zinc also
inhibits the conversion of testosterone to dihydrotestosterone (by
inhibiting the 5-alpha reductase enzyme that catalyzes this conversion).
Low testosterone levels (in males) can occur as a result of SELENIUM
DEFICIENCY.

[7] VITAMIN A helps to regulate plasma testosterone levels (vitamin A
deficiency can lead to sub-optimal plasma testosterone levels).

[8] FENUGREEK SEEDS may increase testosterone levels (steroid saponins
in fenugreek seeds stimulate the release of luteinizing hormone which in
turn stimulates the production of testosterone).

[9] KOREAN GINSENG increases the body's endogenous production of
testosterone.

[10] HORNY GOAT WEED & MACA are claimed to increase testosterone levels.

[11] NETTLE inhibits the binding of testosterone to SHBG, resulting
in lower levels of "bound" testosterone and higher levels of "free"
testosterone (this occurs from Nettle binding to SHBG in place of
testosterone).

[12] SAW PALMETTO inhibits the conversion of testosterone to
dihydrotestosterone (DHT).

[13] TRIBULUS TERRESTRIS is claimed to increase production of
luteinizing hormone which causes the testes to release more
testosterone. Tribulus terrestris is speculated to enhance the
conversion of androstenedione to testosterone (although androstenedione
is the immediate precursor of testosterone, it is speculated that this
conversion occurs under the influence of luteinizing hormone. Research
suggests that daily intake of 750 mg of Tribulus terrestris results in
an increase in free testosterone levels of 30% (only in males) within
five days.

[14] ROYAL JELLY is claimed to contain testosterone.

COMMENT: The following thoughts are my personal opinion. Testosterone
appears to be a safe hormone if potentiated to between 90-100% of normal
free testosterone reference range, especially during strength training
bricks. DIET should be high protein, say 1.7-2.0 grams protein per
kilogram body weight and at least 25-30% fat including meat-sources
saturated fat, which supports the body's natural production of
testosterone. Without such a caloric admixture, the body will not have
the building blocks with which to manufacture testosterone. This means
that during low-fat, low-protein, vegetarian diet intake, the
testosterone levels will be predictably lower than with the opposite
type of diet. Oats, velvet beans, soy isoflavones or soy protein
isolates and garlic should be a staple of the vegetarian diet if
testosterone levels are to be supported. Of the SUPPLEMENTS that may
increase testosterone, saw palmetto, nettle extract, Tribulus
terrestris, fenugreek seeds, modest (not high) dose melatonin, and
acetyl-l-carnitine are recommended. Of the minerals that deficiency may
reduce testosterone levels, there is no need to increase dose above
sufficiency dose. Premium Insurance Caps provides each of the minerals
at max dose in each 14-capsule packet. I do not recommend stimulating
free testosterone levels above 210 pg/ml or above 5% of normal maximum
reference range. By disclosure, some excerpts of this information have
been copied by permission from In-Tele-Health © 2002 (from Hyperhealth
Pro CD-ROM) for educational purposes only. Any and all exogenous
hormonal interventions should be reularly monitored by your healthcare
physician.



-----Original Message-----
From: Forbes-Ewan, Chris [mailto:Chris.Forbes-Ewan@...]
Sent: Monday, 11 October 2004 5:43 PM
To: sportscience@yahoogroups.com
Subject: Lacto-ovo-vegetarian diet and body building


Two years ago I wrote a review paper about the effects of vegetarian eating on
"Effect of Vegetarian Diets on Performance in Strength Sports". It was published
in Sportscience:

(Sportscience 6, sportsci.org/jour/0201/cf-e.htm, 2002).

I recently received a message about an aspect that wasn't addressed in the
review. If anyone would like to attempt an answer to the following question, I'd
be very pleased to forward the answer to the person who sent me the message:

"I have a question that I have been researching and finding minimal answers to.
I am a lacto-ovo vegetarian (no meat and no fish, but I eat eggs and milk),
steroid-free, novice bodybuilder. I have read in numerous books, magazines,
websites, academic papers (such as your own), etc. that vegetarians have lower
testosterone levels than meat-eaters. I have not found a reason why this is the
case. Could it be because in the studies that produce such results, the
vegetarians were not supplementing their diet with enough B-vitamins, zinc,
essential fatty acids, etc. or is there something else inherent in eating animal
flesh that raises one's testosterone level. You mention in your article,
"Deborah Shulman suggested that at least 12 weeks would be needed for studies
comparing the effects on performance at strength sports of nutrient-rich
vegetarian diets with those containing meat." Has a study like this been done in
the past 2 years, or do you think you know what the results woul!
d be? I have considered experimenting with adding meat to my diet for a trial
period simply to see if it has an effect on my bodybuilding and strength
training, despite having been a vegetarian for about 4 years. Do you think it
would be worth it to eat meat? Please tell me any thoughts you have on this
subject. Thanks."


Chris Forbes-Ewan

Defence Scientist (Nutrition) S&T 5
Defence Nutrition
DSTO-Scottsdale
PO Box 147
SCOTTSDALE  Tas  7260
AUSTRALIA

Phone: Int + 61 3 6352 6607 (03 6352 6607 within Australia)
Fax:   Int + 61 3 6352 3044 (03 6352 3044 within Australia)

The opinions expressed in this message are those of the author, and should not
be taken to represent the position of the Defence Science and Technology
Organisation or of the Australian Department of Defence.

IMPORTANT: This email message remains the property of the Australian Defence
Organisation and is subject to the jurisdiction of section 70 of the CRIMES ACT
1914. If you have received this email message in error, you are requested to
contact the author and delete the message.

I believed that the original Tatem paper was published in Nature.  Just like
another paper published in Nature  4 years ago by S Savaglio and V Carbone
'Human performances: Scaling in athletic world records' Nature 404: 244,
2000, the Tatum paper is an example of non-sports science related
investigators playing an interesting, if poorly informed and executed, game
of 'what if'.  Both papers defy common sense, ignore everything we know
about the limitations of human performance and, in my view, hamper the
efforts of serious investigators using modeling techniques to better
understand the determinants (and limits) of human muscular performance.
Granted that Nature doesn't publish very much in the sports science domain
(they have certainly rejected everything I have ever sent to them), if they
are willing to publish such ill-concieved pieces as the Tatem article, one
must wonder about the quality of the other science published in Nature.
Maybe they don't deserve their ~29 impact factor from ISI?  While granting
that paradigm challenges, often from seemingly ill-prepared outsiders, are
the creative energy that drives revolutionary  change in science, the dark
side of such glaring failures of the peer review process is a decrease in
the signal-to-noise ratio relative to really understanding our world.

Carl Foster, Ph.D.
Professor of Exercise and Sport Science
University of Wisconsin-La Crosse

Three years ago I sent a rather ill-conceived message to the Sportscience list
about a supposed narrowing of the gap between elite male and female times for
marathon running.  Stephen Seiler pointed out that the gap between men and women
isn't narrowing, and may actually be increasing slightly. For example, his 1997
article at http://sportsci.org/news/news9705/gengap.html includes:

"Since 1989, world records for males have improved an average of 1.0%, whereas
those for females have improved by 0.0% if we exclude a highly controversial
track meet in Beijing and by 0.3% if we include that meet. Seven of the 10 men's
world records standing in 1989 have been broken in the 90s. Men's world records
set in the 80s had been bettered a total of 23 times by December 1996. In
contrast, women's world records set during or before 1989 have been broken only
once or three times, depending on whether the two controversial Chinese records
are considered legitimate.

"In short, the women are not still gaining on the men. The gap isn't even
stable. The world's fastest female runners have become slower in the 90s! Is the
widening of the gender gap in running "significant"? Statistically, the question
is irrelevant. We are comparing two populations of world elite athletes, not
samples. The increase in the gap is real."

A recent Scientific American article again raises the possibility that, at least
for sprinters, the gap may be narrowing (see the full article below my signature
block). Is this another furphy, or has the situation suddenly and substantially
changed from the late 1990s?

Chris Forbes-Ewan

Defence Scientist (Nutrition) S&T 5
Defence Nutrition
DSTO-Scottsdale
PO Box 147
SCOTTSDALE  Tas  7260
AUSTRALIA

Phone: Int + 61 3 6352 6607 (03 6352 6607 within Australia)
Fax:   Int + 61 3 6352 3044 (03 6352 3044 within Australia)

Scientific American On-Line
September 30, 2004

Data Trends Suggest Women Will Outrun Men in 2156

Every four years, athletes from around the world travel to the Olympic Games to
compete in the 100-meter dash, hoping to earn title of fastest man or woman on
Earth. A new statistical analysis suggests that in the year 2156, the winner of
the women's event may finally outrun her male counterpart.

Andrew J. Tatem of the University of Oxford and his colleagues collected the
finishing times in the men's 100-meter dash run in 1900 and from 1928 (when the
women's race was first run) to 2004. The winning times for both genders have
been steadily decreasing, with female competitors improving at a slightly faster
clip than the males. By plotting the results against the year of competition and
extrapolating the results, the team determined that the fastest human on the
planet could be a woman after the 2156 games. In today's issue of the journal
Nature, they report with a 5 percent margin of error that the event could take
place as soon as 2064 or as late as 2788, however.

Tatem is the first to admit that the study represents a purely academic
exercise. A disease researcher by trade, he says the new study was a result of
noticing a strong and interesting trend in sprinting. Indeed, the relationship
was surprisingly linear and no other model fit the data as well. "We decided to
throw caution to the wind and see if current trends continued, what would happen
in the future," he remarks. Potential confounding factors that are not addressed
in the new analysis include illegal drug use, environmental conditions on race
day, national boycotts and timing accuracy. In addition, some researchers
contend that humans are hurtling toward the limits of their potential and that
the winning times predicted for 2156 (8.079 seconds for the female champion and
8.098 seconds for the male winner) are simply beyond our grasp. The next chance
to check the trend comes in 2008 at Beijing. - Sarah Graham

© 1996-2004 Scientific American, Inc. All rights reserved.

Biostats Group

Joe Weir asked the question below and I figured everyone would have the same
question. You should submit the abstract under Epidemiology. There is no way
to actually indicate that the abstract should be part of the thematic poster
session. The plan is for you to also inform me of the confirmation you
receive (title, etc), and I send the list to ACSM Program committee, and
then they will organize the abstracts into one session.

Hope this clears things up.

Ian Shrier

>-----Original Message-----
>From: Weir, Ph.D., Joseph [mailto:Joseph.Weir@...]
>Sent: 13-Oct-2004 3:37 PM
>To: Ian Shrier
>Subject: RE: ACSM 2005 Biostats Poster Session
>
>
>Ian - what sort of info does the abstract submission have to have in it
>in order to let ACSM know that the submission needs to go to the stats
>session?
>
>Joe
>
>Joseph P. Weir, Ph.D. FACSM
>Professor and Research Coordinator
>Division of Physical Therapy
>Des Moines University - Osteopathic Medical Center
>3200 Grand Avenue
>Des Moines, IA  50312
>515.271.1733
>FAX 515.271.1714
>joseph.weir@...
>www.dmu.edu

Howdy Chris,
     I recommend contacting D. Enette Larson PhD, RD at  the Pennington
Biomedical Research Center, Louisiana State University.  Information on her
work:
http://www.pbrc.edu/facultyprofile.asp?links=research&EmployeeID=1396
     Her e-mail is: larsonde@...
     She is considered "the authority" on nutrition  for vegetarian athletes
in the US.

Ellen Coleman, RD, MA, MPH
Sports dietitian

Biostats Group

Abstract deadline is less than 3 wks away and we need to firm up if there
will be Thematic Poster session. We asked about intentions before, and now
we need some committment. The following were the names and tentative
subjects. If you are on the list, please confirm that you still intend to
submit this. If you have an idea and are not on the list, please let us
know. It would be great to have as many as possible. One session would hold
5-6 abstracts, but who knows, maybe we can get two sessions!

Joe Weir: traditional tests for stationarity (runs test and reverse
arrangements) are not valid indices of stationarity in the frequency domain,
only the time domain

Joel Cramer: The purpose of the presentation would be a "validation" of the
continuous wavelet transform

Dwight The: 1) Fitting the four parameter Generalized Lambda Distribution
(GLD) to exercise science data: Learning what two parameter distributions
(e.g., the Normal) simply cannot tell us OR 2) Invoking different
statistical distributions to (Re)interpret the ACSM Guidelines' "Physical
Fitness Testing Norms"

Gavin Sandercock: Meta-analysis

Munish Chander: Sport injuries


Ian Shrier and Will Hopkins

The following positions were recently posted to HigherEdJobs.com
http://www.HigherEdJobs.com.  We hope this is helpful!  Amy

Longwood University
Lectureship, Health, Recreation, and Kinesiology
Application Due: Open Until Filled
Posted: 10/11/2004
Location: VA
Type: Full Time
http://www.higheredjobs.com/details.cfm?jobcode=175107696

Springfield College
Assistant Professor - Sport Management
Application Due: Open Until Filled
Posted: 10/06/2004
Location: MA
Type: Full Time
http://www.higheredjobs.com/details.cfm?jobcode=175107062

University of Kansas - School of Education
Assistant Professor in Sport Management/Administration
Application Due: Open Until Filled
Posted: 10/04/2004
Location: KS
Type: Full Time
http://www.higheredjobs.com/details.cfm?jobcode=175106706

Grays Harbor College
Athletic Director/Instructor
Application Due: 10/20/2004
Posted: 09/30/2004
Location: WA
Type: Full Time
http://www.higheredjobs.com/details.cfm?jobcode=175106382

California State University, Fresno
Athletic Training
Application Due: Open Until Filled
Posted: 09/28/2004
Location: CA
Type: Full Time
http://www.higheredjobs.com/details.cfm?jobcode=175106050

Two years ago I wrote a review paper about the effects of vegetarian eating on
"Effect of Vegetarian Diets on Performance in Strength Sports". It was published
in Sportscience:

(Sportscience 6, sportsci.org/jour/0201/cf-e.htm, 2002).

I recently received a message about an aspect that wasn't addressed in the
review. If anyone would like to attempt an answer to the following question, I'd
be very pleased to forward the answer to the person who sent me the message:

"I have a question that I have been researching and finding minimal answers to.
I am a lacto-ovo vegetarian (no meat and no fish, but I eat eggs and milk),
steroid-free, novice bodybuilder. I have read in numerous books, magazines,
websites, academic papers (such as your own), etc. that vegetarians have lower
testosterone levels than meat-eaters. I have not found a reason why this is the
case. Could it be because in the studies that produce such results, the
vegetarians were not supplementing their diet with enough B-vitamins, zinc,
essential fatty acids, etc. or is there something else inherent in eating animal
flesh that raises one's testosterone level. You mention in your article,
"Deborah Shulman suggested that at least 12 weeks would be needed for studies
comparing the effects on performance at strength sports of nutrient-rich
vegetarian diets with those containing meat." Has a study like this been done in
the past 2 years, or do you think you know what the results woul!
d be? I have considered experimenting with adding meat to my diet for a trial
period simply to see if it has an effect on my bodybuilding and strength
training, despite having been a vegetarian for about 4 years. Do you think it
would be worth it to eat meat? Please tell me any thoughts you have on this
subject. Thanks."


Chris Forbes-Ewan

Defence Scientist (Nutrition) S&T 5
Defence Nutrition
DSTO-Scottsdale
PO Box 147
SCOTTSDALE  Tas  7260
AUSTRALIA

Phone: Int + 61 3 6352 6607 (03 6352 6607 within Australia)
Fax:   Int + 61 3 6352 3044 (03 6352 3044 within Australia)

The opinions expressed in this message are those of the author, and should not
be taken to represent the position of the Defence Science and Technology
Organisation or of the Australian Department of Defence.

IMPORTANT: This email message remains the property of the Australian Defence
Organisation and is subject to the jurisdiction of section 70 of the CRIMES ACT
1914. If you have received this email message in error, you are requested to
contact the author and delete the message.

Dear List Member,

We are pleased to announce the release of Athletemonitoring.com, a
complete Internet-based athlete monitoring system that helps coaches,
researchers, conditioning specialists and sports medicine
professionals to implement and run comprehensive and practical
athlete-monitoring programs.

Athletemonitoring.com can be customized according to sport and
coaching preferences and is designed to easily:

-      Collect subjective and objective training/competition feedback
-      Monitor well-being, fatigue and mood states
-      Track injuries and self-administered test results/progression
-      Communicate with athletes no matter their geographical location

With Athletemonitoring.com, athletes enter information using their
Internet browser and confidential password. Once information is
entered, coaches can then review and compare training volumes,
competition feedback, evaluate training responses and adaptation to
training, control training attendance during unsupervised sessions,
identify tired athletes, check injury status, identify the causes of
injuries and print/display individual or group-based charts, reports
and statistics.

Athletemonitoring.com is a subscription-based service. Pricing starts
at 1$/month/athlete. An Intranet version is also available for larger
clubs, universities, national governing body of sports, and other
entities that prefer to run Athletemonitoring.com over an internal
network or "intranet".

For more details, please visit http://www.athletemonitoring.com or
email us at info@...

Thank you for your time and attention.

François Gazzano
President

____________________________________
Advanced Fitness Designs, Inc
75 Brydges St.
Moncton, NB, Canada
E1C 2E9

Tel: +1-506-386-1023
Fax: +1-832-565-5976
http://www.athletemonitoring.com
http://www.af-d.com
____________________________________

I sent this to Joe Weir a few days ago, but it bounced back at me.
If anyone on the list is at Joe's institution, please let him know
something's wrong with his mail.


Gidday, Joe.  I fear the problem will be intractable, because you
will need to take into account all the covariances among the
predictors.  I checked to see whether Excel provides new-prediction
errors in its LINEST function, but alas it doesn't.  The closest it
gets is the standard error of the estimate, which of course is the
prediction error only for the observations that produced the
prediction equation.  A package like SAS provides new prediction
errors, but using matrix algebra.  You get the prediction error for
any given new point by including the point in the data set, with a
missing value for the dependent variable.  Output the predicteds and
there you will find the predicted value with its standard error.  You
also get the standard error and confidence limits for the population
predicted value for that point.

Will

Has anybody exerienced any differences between HRs observerd during
an incremental test and during a steady state exercise bout for a
given sub-maximal work rate. Eg, HR during incremental test stage =
150bpm vs HR during 15min bout of work at the same intensity = 128
mean (135 max).

Any thoughts?

Rob

I am looking for a Latin American expert to consult with coaches of
professional baseball in the United States on steroids.  Please
respond to me directly, Michael Barnes, NSCA Director of Eduation at
mbarnes@...

Thank you,
Mike

POST DOCTORAL FELLOWSHIP (3 yrs)

Position Summary

Selection Criteria

Essential

• A PhD, preferably in Exercise physiology
• Experience with conduction of complex human exercise experiments.
• A through understanding of exercise physiology.
• An ability to work very well with people in terms of participant recruitment for human research and ability to retain recruited participants.
• Evidence of independence in the conduct of research.
• Excellent oral and written communication skills

• Self-motivated
• Ability to work both as an individual and a team member
• Ability to maintain accurate records of laboratory activities
• Ability to meet strict guidelines
• Well developed problem solving skills.
• Computer literate in Microsoft Office and statistical software packages