ABSTRACT
The chapter discusses some special features of forest and environmental experiments. It is emphasized that the target population should be clearly defined. The generalization of the results of forest experiments is often problematic because there are many nuisance factors related to variation among people involved, heterogeneity of forests and variation of weather conditions. It is usually not possible to sample treatment units randomly from the target population. Forest experiments are often long term and multipurpose experiments. When the climate is changing, it is difficult to separate the effects of age and climate change. Blocking is an important method for reducing residual variance. It is discussed whether block effects should be random or fixed. The analysis should aim at quantitative models not at obtaining significant p-values. The treatment units and measurement units should not be confused in the analysis of experiments. The intended model should be taken into account in the design of the experiments. Demonstration plots are often important in suggesting hypotheses. The lack of sufficient number of replicates is often problem in forest experiments.
