Dear Will,

I read with interest your report on the ACSM conference. In describing
the talk that I gave there on the analysis of data from mobile cycle
ergometers, you refer to the constants used in the
calculation of normalized power as "arbitrary". In fact, my decision to
apply a 30 s rolling average and 4th order weighting to normalize data
from on-bike power meters was not at all arbitrary, but instead was
based on a wealth of data regarding the time course and non-linear
nature of a wide variety of physiological (especially metabolic)
responses to changes in exercise intensity. Moreover, while the notion
that perhaps individualizing these values might yield a more accurate
adjustment, the fact that normalized power varies by only ~5% when the
constants are varied by 50% (i.e., from 15 to 45 s, or from 2 to 4 for
the exponent) demonstrates that the algorithm is rather insensitive to
the precise values used.

Andy Coggan