Measurement Forum Archive

Read-only historical archive of measure.infopop.cc

Validations 1987-2006

#1 • 9/17/07 @ 2:56 PM
VALIDATION SUMMARY 1987-2006

The USATF/RRTC validation program has been in effect for over two decades. The first course for which we have a record was measured in 1980. Since then 408 courses have been validated (to end of 2006). Validation reports similar to this have been presented in Measurement News , the last appearing in Issue #104, November 2000.

There is an Excel file entitled “val.xls.” It contains information generated by all past Validations Chairpersons, and each time a new report is completed the information is added to the file. The file may be obtained from Pete Riegel (riegelpete@aol.com) for anyone wishing to check the accuracy of this report or do some different research.

The validation data compares the measured length obtained by a skilled and experienced validator with the nominal length of a course which was previously certified and on which a record was set.

Results of my analysis of the data are shown below:



In former times a 10 km course was considered to “pass” validation if it remeasured to not less than 9995 meters, with other distances proportional. This was known as the “allowance for error in the validation measurement” or AEVM. This year the “pass” line was set at 10,000, and the AEVM eliminated. For this reason the analytical results are shown both with and without the AEVM.

I chose the last two decades for the analytical period, as during the earlier years the modern layout procedure was young and not fully developed. For courses certified from 1987 to 2006 we see both the effect of the AEVM and the effect of experience in measurement. While our overall “pass” rate is now 87.5 percent, those courses measured by certifiers amount to 44 percent of the total, with a “pass” rate of 95.5 percent (107 of 112 passing). Those courses measured by others pass at the rate of 81.1 percent (116 0f 143 passing).

The measurements have a fairly normal distribution centered at just over the nominal length plus the short course prevention factor, as would be expected. See below



"Range" above is 1 m/km wide. For example, 15 courses remeasured between -1 and 0 m/km.

Commentary is invited.
#2 • 9/18/07 @ 10:00 AM
It's my understanding that the only time a course is validated is if there is an American record set.

Does that mean someone set an American record on a 10k course that was 270 meters long!!
#3 • 9/18/07 @ 11:19 AM
Mark,

That's right. It was almost certainly an age-group record. It was the 1998 Detroit Free Press Marathon.

It remeasured to 43299.78 meters
#4 • 9/19/07 @ 9:10 AM
Pete,

It would be interesting to pull out the validations that failed because the course described on the certification map was not followed on race day. Then you'd have a better comparison between what the original measurer got and what the validator got.
#5 • 9/19/07 @ 2:19 PM
It would be interesting but I don't have that data. In the file "val.xls" there are notations where the course as run differed from the one certified. This information is probably imperfect. Still, I think it is good enough to learn from.

I think it is safe to say that anything way out beyond the middle is interesting, but not really representative of good comparative measurements, as some sort of gross error occurred.
#6 • 9/24/07 @ 2:48 AM
I've been informed that my comments on why this validation were way off are inaccurate, so I am confessing ignorance. The disparity between the measurments remains a mystery.