rmba wrote:
How did you choose where to start and end your 2 linear regressions? ie was there overlap of the selections? Were you guided by your knowledge of 'about where your threshold lies' currently? I also think the type of regression used is important. I am sure you are aware that simple linear regression is not appropriate when there can be some degree of error of both x and y axes. I don't know without seeing the data but probably Deming regression (or maybe Passing-Bablok) would be appropriate.
These are important since if you got entirely different results, the premise of your question may change entirely.
Along these lines, when I eyeball your plot, running an imaginary line through the middle of the lower power data points (your regression line doesn't appear to run through the middle of this as alluded to by the first response) and another through the higher power data points, the crossover is considerably lower (potentially under 200W).
....in which case (and I may not be correct since I haven't seen the raw data and I am making eyeball assessments), *this* physiological testing measure is probably not useful (since you can probably tell from your workouts that your other estimates of 220-225W are about correct).
I am also not sure what the comparison is to, you say 'performance', have you done a recent 1hr test? (or are you comparing to other 'estimates'?).
Ultimately, *if* you come up with FTP= 220-230W from different approaches (and nice if they back each other up), there is most likely no practical difference where in that range you choose (ie FTP estimates are just that: 'estimates', and, as long as you are close, the precise number should not matter).
Randolph
I minimized the overall sums-of-squares while requiring that the two lines intersect. IOW, tbe standard approach used in the literature for objectively identifying breapoints.
Now as for the difference between what this analysis shows vs. an eyeball test, I'd have to agree with you...but I didn't want to bias the analysis by, e.g., excluding the cloud of points at the lower intensities (before the shifta).
Finally, finding that the NIRS breakpoint corresponds to FTP really isn't surprising, as it has previously been shown to correspond to maximal lactate steady state. A limitation of all such research, though, is that NIRS samples from a rather small tissue volume, and hence from only one muscle, and studies using multiple monitors show that the breakpoint intensity can vary in different muscle groups (reflecting differences in recruitment).