Skip to main content

7 replies

Variation in multiple measurements

R robfrenchMember
Last weekend I measured out my first course. It's a one mile "lollipop" configuration, and I followed the advice posted in this discussion:

https://measure.infopop.cc/eve/...311047343#5311047343

I am considering measuring again, start and finishing with the marks I laid down a week ago, not because I have to, or because I have doubts about the measurement, but because I am curious about the repeatability of measurements. And perhaps I'm overthinking things, but I find it easy to get wrapped up in the numbers and the specification of the methodology.

I work in a world of motor vehicle fuel economy and emissions testing, data collection, and analysis, and I (and the auto companies) live and die by accuracy, repeatability, precision, and rounding (e.g., as a general rule, intermediate rounding is frowned upon unless there is a specific technical reason for doing so). My preference, for example, is to use ASTM rounding when rounding comes into play, but the manual doesn't specify a rounding protocol, meaning that someone else might take the same values and get different results. I also worry about such things as displaying results with a false sense of precision, e.g., are our techniques actually able to measure a 5,000 meter course to the millimeter?

I have searched on a number of relevant terms, and read much about these issues, answering most of the questions that I have had. However, the one thing I am curious about is the measurement to measurement variability in race courses. There's much discussion about variability and required or desired tolerances in the back-to-back calibration taping, or calibration rides, or course measurements performed on the same day.

What I am interested in knowing is the degree to which one certification-worthy measurement done on one day might differ from another measurement effort on another day. In other words, pretend the second measurement is a "confirmation" measurement. Were I to re-measure, I fully expect that I will not land at exactly the same place. I'd be surprised if I was within 1 cm, but this is why I'm asking you folks, since I don't have the breadth of measurement experience. What kind of variability would you expect to see?

Thanks!
Original Post

Replies sorted oldest to newest

genenewmangenenewmanRRTC Chairman
Rob,

Happy to see we will have a new measurer.

You are correct, if you measured the same course on a different day it's highly unlikely you will get the same results. This process actual takes place when we need to check a course for records.

Basically, here is what is done. Let's say it is a 10 km course. After verifying the calibration course is accurate. A person calibrates their bike on this calibration course by taking 4 rides. Next, the course is measured and the counts are recorded. Finally, this person does a post Calibration of 4 rides. The average of the pre and post calibration is found and then without using the .1% adjustment the number of counts needed for this 10 km is found. If the ride had more counts than with this number, the course would pass verification. Note, with the .1% used a 10 km course should be 10,010 meters(this is if we could do a perfect measurement, which is highly unlikely).

As for rounded, I never round until the end. If I where to place mile splits, then I would always round up.
Here is an example of my rounding!

Let's say we have 11,609.3561 cnts/km or 18,683.4476 cnts/mi with the SCPF included. Now we have a course to measure that is a 5 km. I would use 18,684 cnts/mi for each mile. However, I would use 11,609.3561 cnts/km to get total counts for the 5 km. Below is an example.

Measured splits
Start for counter is 1,4445
One mile reading is  1,445 +18,684 = 20,129
Two mile reading is  20,129 + 18,684 = 38,813
Three mile reading is  38,813 + 18,664 = 57,497
5 km point is found as follows: 11,609.3561 x 5 = 58,046.7805(I would always round this up to the nearest whole number).
5 km point reading is 1,445 + 58,047 = 59,492
Note: don’t do the following: 11610 x 5 = 58050+ 1445 or 59,495

Doing it this way makes the course long also, but only by less than a click no matter what the distance is.

Hope this helps! Thanks for your questions.
peteriegelpeteriegelRCMBB Founder
Past issues of Measurement News have discussed this a lot.

To read them go to:

www.rrtc.net

Click on: Historical Archive on RunScore.com

Click on: Measurement News

Click on: Index

Issues 38 and 40 are of special interest, but there is more if you look at the index.

I just found a few more issues with validation info. They are 40, 46, 54, 60, 67, 104.

The online index is not searchable, and I plan to make all issues of Measurement News, with searchable index and all 133 issues, available on the www.jonescounter.com site. All on one DVD - all 520 MB. Watch this space.

If you are serious about your measuring, you will find the historical trail a fascinating one to follow.

Did you know that in the early days a calibration course of 1/2 mile or 1 km was required? If we had not reduced its length, how many courses would we be measuring now?
Last edited by peteriegel
R robfrenchMember
Thanks Gene, this helps.

I run into your final point about what not to do frequently in many different contexts and definitely understand the impact. Rounding and order of operations matters!

Assuming I do go re-measure, mostly for experience and my thirst for data, my inclination would be to use the second set of marks rather than the first. Normally, all other things being equal, there would be no technical reason to pick one over the other, as neither one is likely to be more accurate. But, with another measurement my experience will have increased by 100%, which might mean a more accurate result. Or, to be safe, I could choose to submit for certification the longer of the two.

There are two sources of error/judgment that have bothered me, and your input would be welcome.

The first is in measuring the calibration course with a steel tape. Let's say you have a 100 foot tape with 1/10 and 1/100 graduations, as I do. You stretch the tape and find that your pen mark on the masking tape lies between two 1/100 marks, e.g., between 77.63 and 77.64 feet. What do you do? Do we minimize the error by recording it as 77.635? Or do we use our judgment and guess that it's actually 77.633 or 77.636, which seems to me to be stretching our visual and mental capability.

The second issue is really the same thing, but with partial counts on the Jones Counter. Is it acceptable to submit measurements to the nearest half count, or is that met with raised eyebrows? (I did find a discussion here:

https://measure.infopop.cc/eve/...461097992#2461097992

where Mark Neal suggested that we could read accurately to the 0.5 counts, and Duane Russell suggesting reading accuracy of 0.25 clicks, so I guess I'm looking for confirmation of what is desirable and what is acceptable.) (Your example kind of gets to this by rounding up to get the 5km mark, acknowledging that we have no way of "seeing" a 0.7805 click.)

Thanks for joining me on my foray down into the weeds!

Rob
B bobthurstonMember
Rob, here is what a professional surveyor taught me about using a tape like yours, marked in tenths and hundredths of a foot: Read to nearest hundredth then interpolate to the thousandths place. 1/100 of a foot is a pretty big interval, and reading to the thousandth is analogous to not rounding to the end in your other calculations.

When using a metric tape, you have millimeter marks. I would NOT advise trying to interpolate to the 1/10 millimeter!

To get a feel for how accurate you can reasonably expect your measurements to be, I have 2 suggestions:

1. Make multiple rides of your calibration course, just to see how variable (or not) those rides are. If you want to test accuracy even more, take a magic marker and number your spokes (usually 0 through 35 on a wheel with 36 spokes), then read those "spoke measurements", interpolating between spokes. You will have numbers like 3.6, 14.2, etc.

2. Make repeated measurements of the course, as you intend to do. For a few years when I started out, I followed this suggestion from the fellow who mentored me:
-- first an overall measurement to be able to set start and finish.
-- second, mark mile or other intermediate points.
-- rides #3 and #4 for measuring between those points established in ride 2.
-- for each interval of the course, use whichever measurement finds the given interval shortest, i.e. makes the course longer.
I no longer do 4 rides for most courses, but it was a great way to learn a lot about how to get good tight measurements.

By the way, on the question of which ride to use, don't decide beforehand, but let your results tell you-- use the measurement(s) that result in longest course.

I'd like to add that the above suggestions are NOT part of the general instructions for measuring-- we don't want folks to think they have to get that elaborate just to measure a course! Just mentioning to you because you are interested in those questions of accuracy, repeatability, and precision.
Last edited by bobthurston

Add Reply

Post

Membership Required

We're sorry. You must be signed in to continue. Sign In or Register
×
Are you sure you want to remove from your Block List?
×
Loading...
×
When you block a person, they can no longer invite you to a private message or post to your profile wall. Replies and comments they make will be collapsed/hidden by default. Finally, you'll never receive email notifications about content they create or likes they designate for your content.
Note: if you proceed, you will no longer be following .
×
Link copied to your clipboard.
×