kileyay wrote:
This is a question I've been thinking about a lot. What is the difference between a "source of error" I need to just explain and one I should quantify. In other words, what is industry best practice for error bars here? I can calculate SEM a number of different ways.
This is something that has bothered me about most error bars I see on bike measurements, in the couple places that actually do it (I forget if this applies to the tririg testing), but when they just apply a round number percentage to all the results. First I doubt that actually uncertainty in a measurement would be a nice round number like 3%, because there are so many measurements that go into calculating drag, not just the scale, but also wind speed, temperature, pressure, etc. No way they sum up to a round number. Second the uncertainty is likely not the same percentage for every value. If you have scale that is +/-1% full scale accuracy, the uncertainty percentage of the measurement is going to be different for when measuring 500 grams versus 1000 grams. Simply if bike A has 10% more drag as measured in your test than bike B, the error bars on each should probably not be the same size. They may be similar size if something like the repeat ability of you position is much a bigger factor than the rest of the measurements, but I suspect it is not. I assume you guys have all the specs for the measurements you guys recorded and need to calculate drag?
All I know is testing where I have had to include uncertainty in the data, that has taken more time calculating than the rest of the results. Which makes sense when you start getting into it.
I am very excited to get to read about your results, but boy have you guys picked a very tricky thing to do. The majority of the work and time in this project is in front of you and the fun part is probably behind you, so I do not envy you guys.