LJ and Wind?

[lonewolf]

Someone brought up that you could close or far from the board, but that is essentially irrelevant unless the effect of the wind is different if you are 12cm from the line than if you are 2cm from the line. And while your estimates might be better if you could include exactly how close the the line you are, if you have enough measurements without bias it matters little.

Takeoff point will not affect the distance of the jump ( ignoring the possibility of difference in traction between board and runway) but it will affect the distance measured. Unless you know the exact distance of each jump you are just compiling numbers.
A tail wind helps but unless all factors are identical, and they never will be, you cannot, imo, assign it a meaningful number.

[lonewolf]

More to the point. What is the point of guesstimating that a given jump would have been greater, or less, giving different conditions?
We know a tail wind helps. That is why rules provide for a maximum allowable for records,
Although futile, it is not unreasonable to estimate a given jump would have been recorded as longer if the jumper had not taken before the board but that does not affect the results.

[proofs in the pudd'in]

Given the many variables why not just list the indoor jumps with the outdoor jumps as part of all LJs or don't designate as such just have a legal LJ list? Is a no wind jump really that big of a difference to be noted as indoor as opposed to outdoor particularly given that the outdoor list gives you an idea of wind influence.

[26mi235]

All of the comments here would basically hold for the 100m (reaction time differences are like portion of the board) and most of us accept the utility of the effects of wind on the 100m times.

We know that wind has an effect (if not, then why are jumps over 2.0mps no legal for records). So, the question is what is the gradient. I think everyone agrees that it is not zero. So, yes there are other factors but basically the statistical techniques can control for them to greater and lesser extents, especially given enough data and care in modeling it.

[cajun5150]

Then clearly it is 4 inches!

[Dave]

I think altitude (say for instance Mexico City) and wind (say for instance 2mps) has to be worth at least 1 ft.

Mexico City was also about 8000' so altitude was also helping.

[26mi235]

And it seems likely (90% likely?) that the wind was over 2mps, so assume that it was 2.5, not 2.0 (there was a long thread on this and the mistakes that the guy doing the wind for the LJ (and some other events) was making when the guy who had to often be doing the hurdles was not doing the LJ. The error was truncating all the fractions down so that everything from 2.0 to 2.99 mps was recorded as 2.0 (and the guy's readings were virtually always an integer (2.0, 3.0, etc).

[Dave]

This is an interesting concept to use a statistical approach rather than some sort of calculated method.

what would you do? You would take all the jumps you can find for a given athlete with windspeed, altitude, and distance. Then feed that into some sort of correlation analysis? I have no idea if you would lump in all available data for multiple athletes. It seems more valuable to look at all recorded jumps by a given athlete rather than trying to compare multiple athletes.

Hopefully, we have a statistician available.

[lonewolf]

It would make more sense to analyze one athlete than many.

[gh]

the only methodology that's acceptable is a tough one, because you need somebody to take a shitload of jumps, starting with (as RV Hill did with the research that lead to the T&FN charts) extensive wind-tunnel numbers.

Same person(s) need to jump "many" times in wind-zero conditions (obviously indoors), then jump the same large number with wind-tunnel setup and the fan generating various speeds. Any "statistical" analysis simply won't mean much because there won't be enough same-situation data available.

[lonewolf]

Sounds good except jumper will tire and there is no assurance or possibility that he/she will put same/equal effort into every jump..and, if this experiment eliminates fouls and measures actual distance of each jump, results will be positively flawed because jumper can jump with abandon.

[26mi235]

There is not need to have a bunch of wind-zero jumps, although that is somewhat useful. A large amount of data are used in a method that uses "fixed effects" as well as variable effects. A simple fixed effects model applied to the long jump might have a fixed effect for a location or a meet (and use a number of locations, the jumper) and use variations in wind to capture the differential effects of wind. The wind variable would be estimating the gradient across a large number of observations while the various fixed effects factors helps control the other variables.

As opposed to a simple multi-variable regression analysis using a fixed-effects type model allows for using correlations across a large number of observations. Once you start to have thousands of observations and the correlation structure to, potentially provide added precision you might be surprised at the degree to which you can get estimates of the effects.

Anyway, that is where I would start if I were estimating the impact of wind on the LJ.

[gh]

you'd be starting in the wrong place, sorry

[26mi235]

you'd be starting in the wrong place, sorry

Not sure what you mean.