Effect of lane allocation

[loverarge]

I am currently doing a statistical project, where i intend to investigate the effect on times of running in the different lanes, to see if there is any real difference between the 8 lanes. To do this i need the World Records for each individual lane.

http://users.rcn.com/bricklan/athletic/lanes/lanes.html

The website above gives this but it dated 2003, so i need to update this.

If anyone is able to help me in updating this it would be much appreciated.

From 17/10/2003 I have the world records for each lane as follows:

MENS 200m:
1 Coby Miller USA 20.04 1 Athl Lausanne 1Jul2003

2 Robson da Silva BRA 20.00 1 WCup Barcelona 10Sep89

3 Michael Johnson USA 19.32 1 OG Atlanta 1Aug96

4 Pietro Mennea ITA 19.72A 1 WUG Mexico City 12Sep79
Michael Johnson USA 19.79 1 WCh Goteborg 11Aug95

5 Michael Johnson USA 19.66 1 FOT Atlanta 23Jun96

6 Joe DeLoach USA 19.75 1 OG Seoul 28Sep88

7 Carl Lewis USA 19.80 1 OG Los Angeles 8Aug84

8 Michael Johnson USA 19.79 1 FOT New Orleans 28Jun92

MENS 400m:
1 Mark Richardson GBR 44.37 1 Bislett Oslo 9Jul98

2 Larry James USA 43.97A 2 OG Mexico City 18Oct68
Michael Johnson USA 44.17 1r1 Athl Lausanne 10Jul91

3 Michael Johnson USA 43.65 1 WCh Stuttgart 17Aug93

4 Butch Reynolds USA 43.29 1 WkZ Zurich 17Aug88

5 Michael Johnson USA 43.18 1 WCh Seville 26Aug99

6 Michael Johnson USA 43.66 1 USATAF Sacramento 16Jun95

7 Butch Reynolds USA 44.13 2 WCh Stuttgart 17Aug93

8 Antonio Pettigrew USA 44.45 5 USATAF Eugene 19Jun93

9 Butch Reynolds USA 44.43 1h1 NCAA Baton Rouge 4Jun87

MENS 400m HURDLES:
1 Angelo Taylor USA 47.50 1 OG Sydney 27Sep00

2 Edwin Moses USA 47.45 1 AAU Westwood 11Jun77

3 Brian Bronson USA 47.03 1 USATF New Orleans 21Jun98

4 Kevin Young USA 46.78 1 OG Barcelona 6Aug92

5 Edwin Moses USA 47.02 1 Invit Koblenz 31Aug83

6 Edwin Moses USA 47.13 1 Lanz Milan 3Jul80

7 Danny Harris USA 47.63 1 WkZ Zurich 21Aug85

8 Edwin Moses USA 47.58 1s1 FOT Los Angeles 17Jun84

9 Edwin Moses USA 47.37 1 WCup Rome 4Sep81

WOMEN 200m:

1 Marita Koch GDR 21.90 1 WCup Canberra 4Oct85

2 Marita Koch GDR 21.71 1 v Can Karl-Marx-Stadt 10Jun79

3 Heike Drechsler GDR 21.71 1 GDR Ch Jena 29Jun86

4 Merlene Ottey JAM 21.64 1 VDamme Brussels 13Sep91

5 Florence Griffith-Joyner USA 21.34 1 OG Seoul 29Sep88

6 Grace Jackson JAM 21.72 2 OG Seoul 29Sep88

7 Valerie Brisco USA 21.81 1 OG Los Angeles 9Aug84

8 Evelyn Ashford USA 21.83 1 WCup Montreal 24Aug79

9 Marion Jones USA 21.62A 1 WCup Johannesburg 11Sep98

WOMEN 400m:
1 Olga Vladykina UKR 48.27 2 WCup Canberra 6Oct85

2 Marita Koch GDR 47.60 1 WCup Canberra 6Oct85

3 Jarmila Kratochvilova CZE 47.99 1 WCh Helsinki 10Aug83

4 Cathy Freeman AUS 48.63 2 OG Atlanta 29Jul96

5 Valerie Brisco USA 48.83 1 OG Los Angeles 6Aug84
Marie-Jose Perec FRA 48.83 1 OG Barcelona 5Aug92

6 Jarmila Kratochvilova CZE 48.61 1 WCup Rome 6Sep81

7 Chandra Cheeseborough USA 49.05 2 OG Los Angeles 6Aug84

8 Marita Koch GDR 48.60 1 ECup Turin 4Aug79

WOMEN 400m HURDLES:
1 Yulia Pechonkina RUS 53.38 1 ECup Annecy 22Jun02

2 Janeene Vickers USA 53.71 2 Wkz Zurich 7Aug91

3 Tonja Buford USA 52.62 2 WCh Goteborg 11Aug95

4 Yulia Pechonkina RUS 52.34 1 NC Tula 08Aug03

5 Kim Batten USA 52.61 1 WCh Goteborg 11Aug95

6 Sandra Farmer-Patrick USA 52.79 2 WCh Stuttgart 19Aug93

7 Kim Batten USA 53.17A 3 WCup Johannesburg 11Sep98
Sandra Cummings Glover USA 53.33 1 USATAF Sacramento 17Jul02

8 Nezha Bidouane MAR 52.96A 1 WCup Johannesburg 11Sep98
Ionela Tirlea ROM 53.48 4 WkZ Zurich 12Aug98

[Jon]

Re: Effect of lane allocation

MENS 200m:
1 Coby Miller USA 20.04 1 Athl Lausanne 1Jul2003

2 Robson da Silva BRA 20.00 1 WCup Barcelona 10Sep89

3 Michael Johnson USA 19.32 1 OG Atlanta 1Aug96

4 Pietro Mennea ITA 19.72A 1 WUG Mexico City 12Sep79
Michael Johnson USA 19.79 1 WCh Goteborg 11Aug95

5 Michael Johnson USA 19.66 1 FOT Atlanta 23Jun96

6 Joe DeLoach USA 19.75 1 OG Seoul 28Sep88

7 Carl Lewis USA 19.80 1 OG Los Angeles 8Aug84

8 Michael Johnson USA 19.79 1 FOT New Orleans 28Jun92

Some updates:
4 Tyson Gay USA 19.76 1 WCh Osaka 30Aug07
5 Usain Bolt JAM 19.19 1 WCh Berlin 20Aug09
6 Usain Bolt JAM 19.57 1 MVD Brussels 4Sep09
8 Xavier Carter USA 19.63 1 Athl Lausanne 11Jul06

[Marlow]

Re: Effect of lane allocation

I am currently doing a statistical project, where i intend to investigate the effect on times of running in the different lanes, to see if there is any real difference between the 8 lanes.

I would think that it's highly dependent upon the individual athlete's ability to run the curve. Some can, some can't. I'm not sure that one could quantify the differences. Clearly Lane 1 is a disadvantage not only because it is the tightest, but often one can't run as close to the edge of the lane as one can in other lanes. Then there's the problem of 'pulling'. Outside lanes have a gentler curve, but there's fewer athletes to 'pull' on (a very real psychological effect).

[scottmitchell74]

Nice points on that Marlow. I find the lane "dilemma" very interesting as well. Be curious what the raw stats say in the end (keeping in mind that final raw stats won't tell the whole story).

[Frans Rutten]

I don't know if this has ever been theoretically studied in a scientific manner. JRM is the authority I would suggest. But I think, that a study based on best-ever runs for every specific lane for a number of reasons can't have much meaning other than a descriptive display of facts.

In order to investigate the effect in practice for say Bolt you have to have 8 or 9 simultaneous executed maximal runs.
Bolt says all the time he can do a lot, but that

[loverarge]

Re: Effect of lane allocation

I would think that it's highly dependent upon the individual athlete's ability to run the curve. Some can, some can't. I'm not sure that one could quantify the differences. Clearly Lane 1 is a disadvantage not only because it is the tightest, but often one can't run as close to the edge of the lane as one can in other lanes. Then there's the problem of 'pulling'. Outside lanes have a gentler curve, but there's fewer athletes to 'pull' on (a very real psychological effect).

A statiscal study has been done before by R. HUGH MORTON. It was published in a Journal 'The Statistician' in 1997.

http://www3.interscience.wiley.com/jour ... 8/abstract

He concluded:
' Apart from obvious direct effects of gender, distance and the imposition of hurdles, world records for the six lane-affected races do not show any consistent pattern. The less curved outer lanes do not appear to have the expected advantage: nor do the inner lanes appear to show any visual advantade as a result of the staggered start. Significantly better times in the central lanes in only two of the six events suggest that allocating the higer seeded runners to lanes 3-6 is only weakly confirmed. It is clear that, although the weight of anecdotal and conventional wisdom may favour seeding, stronger scientific evidence is needed to justify its retention.'

For my university project i need to replicate this to show i am compitent in using these statistical techniques. I then need to look into expanding on his findings.

You say that some athletes can run the curve and some can't. Do you think this could have any relation to the physical attributes. For example there height, weight ect.

[dj]

The problem now is that only the first round of a multi-round event would be randomly drawn.

The better athletes in such a heat would not have to run all out to advance to the next round.

Only in the final would you have a situation in which all runners are (presumably) running at maximum effort, but you have a biased sample because the "best" runners have been seeded into the most desired lanes.

[gh]

The problem, I suspect, is that the original work was done by somebody with little/no understanding of the sport.

The only way I can think of to prove what we all know is empirically true is to take a trained sprinter and over the course of 8 days, having him run 100s around the curve (no need to run a full 200). Each one full out, with multiple iterations.

Do it over the course of 4 hours, with a half-hour rest between each. On Day one the first race is lane 1, on up to 8. On day 2, start with 2, end with 1, etc., etc. So each lane gets a run with the athlete being at all stages of fatigue.

There's a decent science experiment that doesn't really on faulty data analysis. You simply can't use any actual race results.

[Daisy]

Do it over the course of 4 hours, with a half-hour rest between each. On Day one the first race is lane 1, on up to 8. On day 2, start with 2, end with 1, etc., etc. So each lane gets a run with the athlete being at all stages of fatigue.

Clearly someone trained as a scientist. Might be good to randomize it a little too, so that on any given day the runner never knows which lane is the next in the series.

[gh]

And you probably don't to let the athlete know what any of the times are until it's over.

[26mi235]

... having him run 100s around the curve (no need to run a full 200). Each one full out, with multiple iterations.

While seemingly obvious, I am not sure that his is true. Specifically, trying to run the curve very hard (fighting with constant acceleration inward) might lead to fatigue effects later in the race. Conversely, because running faster in the tightest lanes is so hard, some might take a strategy of going just a little easier on the curve and then being able to push harder. I suspect that this is more important in the 400 than the 200 due to the greater effect of fatigue, but the 200 is not completely a full-out blast with the same energy used, per meter, as the 100m.

[DecFan]

There is surely a lot of noise in the system, but there are advanced statistical techniques available to deal with such data. And you have piles of observations . . .
I do think you can - and should - use real race data, finals only, identifying and statistically compensating for each athlete and each location, and consider wind as well. In allowing for the ability to run the curve, you'll have to allow your estimator to calculate different curve effects for different runners, and then test to see if those adjustments improve the fit of the equation or not.

To do this, you will need multiple observations of each athlete you include in the data set, and preferably multiple observations for each athlete running in each lane.

Should be an interesting project.

[2 cents]

This thread conjured up in my mind a little anecdote which is only peripherally related. I had a training mate who was not so durable, and he wanted to do some of our repeats clock-wise as opposed to the usual counter clock-wise. Well, we would start and finish on the usual start and finish line. Consequently, you'd essentially be finishing the repeat 400, for example, on the curve (using the term loosely) instead of the straight. If the 800 and mile were run this way (finishing on the curve), it might have an interesting effect on strategy.

[gh]

There is surely a lot of noise in the system, but there are advanced statistical techniques available to deal with such data. And you have piles of observations . . ....

What you don't have is an athlete of Bolt, Gay, etc (or before them MJ) caliber running in anything but the middle lanes. Unless you know what the same athlete would do in 1 and 8, compared to normal 4-5-6 I don't see how all the data in the world can be brought to bear.

Your "piles of observations" would be fraught with bias from the get go because of the way lanes are assigned to begin with.

Not to mention the fact that using data from different races is going to bring wind into the equation (something that can't adequately be factored out because of the vectoring problem), and the fact that not all curves are of the same radius. Your usable data gets whittle down rather quickly.

[DecFan]

What you don't have is an athlete of Bolt, Gay, etc (or before them MJ) caliber running in anything but the middle lanes.

If you don't have enough lane variability on Bolt and Gay, leave them out. The Shaun Crawfords of the world will provide sufficient information. Athletes numbered 5 to 40 on the annual list for the last 20 years will provide plenty of data.