Re: Statistics (archive)

Jim,
Thanks for the reply, no disrespect taken. My point comes from my scientific background and I'm probably going a bit overboard on this as I fully understand that the dyno numbers are backed with real world experience concerning power and torque increases. I also understand that our common practice is to pick the median curve and use that to show the differences between set-ups. That's just how it's done.

My point regards the folks who are going through this excercise and are trying to eek out the very last bit of power they can from their cars. Potentially, this can come down to very small differences between set-ups, a little more HP here, a little more torque there. We may very well be talking about 3, 2, or even 1 HP differences. And as my partner always drills in to me: It's a question of signal to noise.

The real crux of the my questions concerns the error of the measurement. There are two sources of error: the error factor in the equipment, and the standard error in the test. My point is that without doing some statistical analysis of these factors one could have an error factor that overshadows the comparison between set-ups. Even a simple standard deviation calculation can give a better idea of the spread of the data.

I'm not looking for the averages of 3 or 4 runs, I'm trying to get a handle on the tolerances involved with the dyno test (the "plus or minus" if you may). A lot of the data points across the three runs you posted are within a few HP of each other, if the error was also a few HP, then any comparison becomes very difficult to make. The average and minimum values in the dyno summary aren't important, but the peak value is. And as an example, run 2 (stock) peak HP was 303.8 and run 20 (Dinan) peak was 304.6. If the error was greater than 0.4 HP (maybe not unreasonable) then the data overlaps and you really can't make a comparative judgement between them. (Yeah, yeah, I see that run 12 (Conforti) was a good 10 HP better!)

So...I'm not necessarily looking for an answer here, I'm just trying to understand how the numbers are derived and if the comparisons are believable. Frankly, I wouldn't mind seeing this thread go away, as I think it has caused too much division in the group. I think everyone is looking for honest answers regarding the best equipment for their cars, even if we don't always express it the right way.

Regards to all,
Stan C.

Follow Ups:

[ Follow Ups ] [ 8-series (E31) Message Board ] [ Msg. Board FAQ ]

Questions, comments, or problems, please visit the Roadfly help desk.