Fig. 1. Points (marked with ‘@’) show he relative frequencies of damaged treated seedlings, *f*_{2}, in different stands with respect to the relative frequency of damaged control seedlings, *f*_{1}. The error bars in both directions indicate the standard error computed as . The thin solid lines describe the median and 95% confidence interval when the probability of damage of treated seedlings, *P*_{2i}, is predicted from *f*_{1} using the estimated model and simulation assuming that each stand has four plots with 17 seedlings in each plot. The thick solid lines show the median and 95% confidence interval when *f*_{2} is predicted from *f*_{1}.

Fig. 2. The solid lines show the predicted stand level probability of damage and the 95% confidence interval in treated seedlings as a function of the stand level probability *P*_{1i} of damage in control seedlings when distance to the humus is equal to its overall average. The maximum value for *P*_{1i} is *P*_{1i} = 0.9999995 which is the upper 95% confidence limit for *P*_{1i}. The dashed lines are obtained when the distance to the humus is the average value ± sd. All curves go to one when *P*_{1i} goes to one.