Age-Related Performance Decline

[leadville]

An article - lengthy, well-researched, and provocative - in today's NYTimes explores the use of and arguments for and against testosterone boosters for aging men.

http://www.nytimes.com/2010/01/17/magaz ... wanted=all

[ranger]

An article - lengthy, well-researched, and provocative - in today's NYTimes explores the use of and arguments for and against testosterone boosters for aging men.

http://www.nytimes.com/2010/01/17/magaz ... wanted=all

When you are 60, you don't need to take drugs to lose weight, get in shape, and perform well in sports.

Just work hard and watch what you eat.

This year, I might well get to 9% body fat and pull sub-6:20 on the erg at 160 lbs.

I will be 59 this week.

ranger

[leadville]

Ranger once again displays his ignorance and inability to understand basic science.

Sigh.

[jliddil]

(snip)

I think that the decline shouldn't be considered an average, but looked at by 5 or 10-year spans. From 25-26 years of age (or some similar gap), there will be an increase in wattage if anything.

Paul

I assume you meant 36 years, since that's how old Benton was when setting the year's best time. Conventional wisdom would suggest a drop in times (also increase in watts) up until a year best set somewhere in the low 30s, followed by slowing times/watts decline. True, conventional wisdom would be dead wrong re LWs, where Henrik Stephensen lately has been a good decade ahead of the expected trajectory .....

You're quite right, though, that a simple linear regression is going to miss improvement into the 30s and a delayed decline. If you get creative with the math you can fit a rather better curve. Here's a 6th order polynomial fit:



And you can do better still by restricting yourself to fitting a curve to the top times by age:



This last is actually roughly what I'd expect: a slight improvement in best times from the mid-20s into the early 30s, followed by a general decline continuing through the 40s into middle age....

Uh don't you have to be careful with this "data". Lies damn lies and statistics.

What about a p value? You were not blinded to the data. You already knew the ages vs times.

[bloomp]

A p-value would be nice, I would gladly run in through MatLab if I still had a copy of that program. That'd give a whole new area of statistics to look at...

[jliddil]

A p-value would be nice, I would gladly run in through MatLab if I still had a copy of that program. That'd give a whole new area of statistics to look at...

You could use R

http://www.r-project.org/

[Nosmo]

You don't need a P-value to tell that a sixth order fit to that data is overfitting.

[NavigationHazard]

Absolutely so. I think the main point from dinking around with this data is that cross-sectional analysis of age-related decline is pretty much useless.

The underlying ranking-page population these days is way too heterogenous for anything particularly meaningful to appear. You can sort of make it appear (or seem to appear) if you massage the data enough. But you're manufacturing results rather than reporting them.

I'd say we're left with what we already knew:

1) Other things equal, we should expect to slow down from middle age on*;
2) Where youth ends and middle age begins is a fuzzy boundary;
3) In the short to medium term, other things are not necessarily equal.

Thus we might as well carry on and accept the consequences gracefully.

* 'In the long run, we are all dead.' -- J. M. Keynes

[jliddil]

Absolutely so. I think the main point from dinking around with this data is that cross-sectional analysis of age-related decline is pretty much useless.

The underlying ranking-page population these days is way too heterogenous for anything particularly meaningful to appear. You can sort of make it appear (or seem to appear) if you massage the data enough. But you're manufacturing results rather than reporting them.

I'd say we're left with what we already knew:

1) Other things equal, we should expect to slow down from middle age on*;
2) Where youth ends and middle age begins is a fuzzy boundary;
3) In the short to medium term, other things are not necessarily equal.

Thus we might as well carry on and accept the consequences gracefully.

* 'In the long run, we are all dead.' -- J. M. Keynes

"Hope I die before I get old." The Who
"Better to burn out than it is to rust" Neil Young

[NavigationHazard]

"You're only as old as the scotch you drink. And I propose to stay at least 12 years old as long as I can afford it...." -- attributed to Groucho Marx.

[bloomp]

Absolutely so. I think the main point from dinking around with this data is that cross-sectional analysis of age-related decline is pretty much useless.

The underlying ranking-page population these days is way too heterogenous for anything particularly meaningful to appear. You can sort of make it appear (or seem to appear) if you massage the data enough. But you're manufacturing results rather than reporting them.

I'd say we're left with what we already knew:

1) Other things equal, we should expect to slow down from middle age on*;
2) Where youth ends and middle age begins is a fuzzy boundary;
3) In the short to medium term, other things are not necessarily equal.

Thus we might as well carry on and accept the consequences gracefully.

* 'In the long run, we are all dead.' -- J. M. Keynes

"Hope I die before I get old." The Who
"Better to burn out than it is to rust" Neil Young

Note the quote in my signature... Mr. Young is quite right.

But yes, Nav, it's impossible to create a mathematical formula that states when one athlete will begin to decline or how long another one will keep getting better. Perhaps Mr. Cureton's upcoming performance will speak loads to one or the other.

[Byron Drachman]

Absolutely so. I think the main point from dinking around with this data is that cross-sectional analysis of age-related decline is pretty much useless.

The underlying ranking-page population these days is way too heterogenous for anything particularly meaningful to appear. You can sort of make it appear (or seem to appear) if you massage the data enough. But you're manufacturing results rather than reporting them.

I'd say we're left with what we already knew:

1) Other things equal, we should expect to slow down from middle age on*;
2) Where youth ends and middle age begins is a fuzzy boundary;
3) In the short to medium term, other things are not necessarily equal.

Thus we might as well carry on and accept the consequences gracefully.

* 'In the long run, we are all dead.' -- J. M. Keynes

"Hope I die before I get old." The Who
"Better to burn out than it is to rust" Neil Young

One of my favorite death bed quotes:

"Everybody has got to die, but I have always believed an exception would be made in my case. Now what ?" William Saroyan

[Yankeerunner]

"Enjoy every sandwich."
--Warren Zevon

[joe80]

C2 UK have now organised three sessions on RowPro for the Race Night Series. Raw results are treated to adjustment by age, gender and weight in order to declare a winner. The adjustment tables are here:

http://concept2.co.uk/rns/adjustment

Summarising the age adjustments (all based on extra time awarded per year for a full 2k i.e. not splits):

0.0s - 20-31
0.8s - 32-37
1.2s - 38-43
1.5s - 44-50
2.0s - 51-55
3.0s - 56-60
4.0s - 61-70
5.0s - 71-80
6.0s - 81-90
7.0s - 91-100

These seem generous allowances in the context of the discussion on this thread. I'm not complaining.

I don't have any idea how C2 produced these figures.

The adjustments have not been universally popular. See this thread:

http://concept2.co.uk/forum/viewtopic.php?f=10&t=20248

Regards,

Joe

[ranger]

C2 UK have now organised three sessions on RowPro for the Race Night Series. Raw results are treated to adjustment by age, gender and weight in order to declare a winner. The adjustment tables are here:

http://concept2.co.uk/rns/adjustment

Summarising the age adjustments (all based on extra time awarded per year for a full 2k i.e. not splits):

0.0s - 20-31
0.8s - 32-37
1.2s - 38-43
1.5s - 44-50
2.0s - 51-55
3.0s - 56-60
4.0s - 61-70
5.0s - 71-80
6.0s - 81-90
7.0s - 91-100

These seem generous allowances in the context of the discussion on this thread. I'm not complaining.

I don't have any idea how C2 produced these figures.

The adjustments have not been universally popular. See this thread:

http://concept2.co.uk/forum/viewtopic.php?f=10&t=20248

Regards,

Joe

So next year, I should be 25 seconds slower over 2K than I was in 2001?

Hmm.

I suppose we'll see, but I think I will be quite a bit better.

It depends primarily on how hard you work at it.

Paul Hendershott was quite a bit better at 60 than he was at 55.

ranger