Using our previous data as representative for this study, the power analysis was based on the prediction that treatment will evoke dose-related increases of the Y-intercept of the dose-effect curve relating the effect to the Y-variable. Accordingly, the power analysis was performed in terms of a two-factor ANOVA with two levels. The standard error of the estimate for the effect is 7% based upon a regression analysis of our existing dose-response data (group sizes were > 8). The effect size that we are designing the study to detect is 1.5 times this standard error. Using n= 10 per cell yields a power of 0.98 to detect this effect size.


The power analysis follows:

Power Analysis for ANOVA Designs

The power parameters you specified were:

  • a = ‘4’ (levels of factor for power)
  • b = ‘2’ (levels of factor(s) crossed with A)
  • delta = ‘1.5, 1.75, 2.0’ (effect size(s))
  • alpha = ‘0.05’ (significance level)


Power analysis for ANOVA designs

4x 2 layout Ha: T1=GM-Delta/2, T2=T3=…=T(k-1)=GM, Tk=GM+Delta/2 tested at Alpha= 0.050


DELTA (in units of sigma=Stdev)

N         1.500   1.750   2.000

10           0.982     0.997    0.999