Importance of Height and Weight on 2K Times

[Pie Man]

Hi Ralph,

Rather than considering the 2k time shouldn't you consider power output as time is not a linear relation? At a guess this will have been blunting the effect of the larger heavier rowers. Hopefully this should only be a simple change to your spreadsheet.

[Carl Watts]

I think you guys are having a nice chat to yourselves but I have not done statistics for over 30 years so this is all pretty meaningless to me.

Probably a far better idea to clarify it into something that everyone can get a handle on, e.g. "95% of all Olympic rowing medal winners are over 6"4' tall" then its not open to interpretation.

[Tim K.]

To sum up, the 2002-2003 data indicate that height and weight matter, just not very much.

Just to make sure I understand, a 200 lbs male who is 6" taller than another 200 lbs male, all other things equal would row 3.72 - 7.68 seconds faster in the 2k and 21.72 - 32.16 seconds faster in a 6k, simply because he is 6" taller? or am I misunderstanding?

[Ralph Earle]

To sum up, the 2002-2003 data indicate that height and weight matter, just not very much.

Just to make sure I understand, a 200 lbs male who is 6" taller than another 200 lbs male, all other things equal would row 3.72 - 7.68 seconds faster in the 2k and 21.72 - 32.16 seconds faster in a 6k, simply because he is 6" taller? or am I misunderstanding?

Essentially, Yes, Tim. The "± standard deviation" is the range in which about 2/3rds of the population of men 6"-taller-but-no- heavier are expected to fall.

So your 6" taller 200-lb man has about one chance in six of being more than 7.68 seconds faster but also one chance in six of being less than 3.72 seconds faster over 2K

[Ralph Earle]

Hi Ralph,

Rather than considering the 2k time shouldn't you consider power output as time is not a linear relation? At a guess this will have been blunting the effect of the larger heavier rowers. Hopefully this should only be a simple change to your spreadsheet.

Here are the results using Average Watts as the dependent variable.

2K
Women: n = 192; Inches = +2.16 ± 0.71; Pounds = +0.80 ± 0.09; Rsq = 0.57.
Men: n = 311; Inches = +2.89 ± 1.06; Pounds = +1.50 ± 0.12; Rsq = 0.72.

6K
Women: n = 192; Inches = +3.31 ± 0.55; Pounds = +0.56 ± 0.08; Rsq = 0.56.
Men: n = 311; Inches = +3.50 ± 0.73; Pounds = +0.87 ± 0.09; Rsq = 0.61.

For both 2K and 6K, when the dependent variable is Seconds, additional height is more valuable to women than to men, while additional weight is equally valuable to both men and women. But when the dependent variable is Watts, both height and weight are more valuable to men. This is because men are, on average, faster and Watts, being a cubic function of power, are harder to gain the faster you go.

[left coaster]

Nice, careful what you ask for Pie Man Ralph is on the scene!

Ralph, did you run these variables together in a multiple regression equation?

I'm guessing you don't have other information such as age or body composition? I'd be curios to see an exploratory/forward regression model with the "*** DELETE - SPAM *** sink" of possible IV's thrown in.

Edit: so strange, for some reason "*** DELETE - SPAM ***" is tagged as spam and won't display... did I say *** DELETE - SPAM ***? .... edit#2 well, you know, the place where you prepare dinner which happens to also have a sink lol.

[Ralph Earle]

Yes, both height and weight were in the regressions.

The only other data available is the month of the test result. I included year (2002 vs 2003) in the 6K analyses to account for possible training effects during the season. It was significant for the women (20 seconds faster) but not for the men. All the 2K tests were from Jan-Apr 2003.

I expect that age would be significant, and might boost the Rsq by 0.10 to 0.20.

[left coaster]

I'm not sure what age rowers peak -- do you think age would be more of a quadratic function rather than a straight linear? i.e. if they peak in late 20's for example?

[lindsayh]

I'm not sure what age rowers peak -- do you think age would be more of a quadratic function rather than a straight linear? i.e. if they peak in late 20's for example?

I think a lot of the really elite guys peak in their 30s rather than earlier

[Pie Man]

Thanks Ralph interesting stats, I had a quick look at the raw data and it made me smile to see the second fastest light weight man in April 2003 is 5'8" and 153 pounds.... He is certainly punching above his weight, or everyone else is below.

[Ralph Earle]

I'm not sure what age rowers peak -- do you think age would be more of a quadratic function rather than a straight linear? i.e. if they peak in late 20's for example?

You are correct, Sir! Here is the Nonathlon's Excel equation for predicting the 1000-point 2K time for heavyweight men from their age:

m:ss.0 = 0.000000528591985186883*Age^2 - 0.0000299485040367209*Age + 0.00442761062635057
R² = 0.979

For lightweight men:

m:ss.0 = 0.000000360148633185895*Age^2 - 0.0000173374404796441*Age + 0.00447197521903429
R² = 0.945

For heavyweight women:

m:ss.0 = 0.000000498553178436036*Age^2 - 0.0000219463709052632*Age + 0.0048266609289393
R² = 0.929

For lightweight women:

m:ss.0 = 0.000000541245933000195*Age^2 - 0.0000322411545085713*Age + 0.00540550759183017
R² = 0.897

The most casual observer of the meanest intelligence will have immediately noticed that, comporting with popular wisdom, women are less predictable than men.

The advanced elementary school student will have already differentiated each equation to find its maximum.