D2 offers a number of useful properties compared to I. In contrast to I, D2 reflects the relative variance expansion due to the between trial variance estimate without assuming an estimate of a 'typical' sampling error σ.
The relative variance in variability, lnRθ, can be assessed instead by estimating the relative variance in repeat number, lnRV, or heterozygosity, lnRH, for loci between populations.
However, large discrepancies were observed for all sampling stages of both procedures between absolute theoretical and experimental relative variances.
Figures 2 to 4 illustrate the different relationship of between- to within-subject variance and the corresponding difference between O R and O A. In the case of SHBG, within-subject variance is small relative to between-subject variance.
These F-ratios were significantly smaller in the heterogenized design (F1,28 = 4.678, p = 0.039, Fig. 8), demonstrating that heterogenization did indeed increase within-experiment variation (variance among blocks of the same laboratory) relative to between-experiment variation (variance among blocks of different laboratories).
Attenuation bias reflects high within-subject variance when measuring risk factors (relative to between-subject variance) and is not reduced by increased sample size, as demonstrated in the Framingham Heart Study (Rosner et al. 1992).
We now turn to the relative between-site variance of the stimulus, as measured by the fluctuation amplitude (i.e. the coefficient of variation).
The fractional anisotropy [11] (FA) defined by:
The analysis was performed on unit variance scaled data, assuming equal importance of each protein regardless of their relative abundance and magnitude of variance between the samples.
FST is an index of population differentiation, measuring the variance between subpopulations relative to the total variance, and ranges from 0 (no divergence) to 1 (complete divergence).
