Which model/statistical test should I use?


Soon I will be performing an experiment where I must grow 4 different plant species under 4 different stressors, each stressor containing 4 levels of 'severity' of the stressor or in other words a difference in concentration. Growth of the plants will be measured over 22 time points, but all these time points will be reduced to a single value per plant, showing the overall growth of the plant over the course of the experiment.

So each observation contains the following data:
  • 1 value giving its growth (continuous variable)
  • 1 value giving the plant species (discrete variable)
  • 1 value for the stressor (type of stressor) (discrete variable)
  • 1 value for the gradient step of the stressor (discrete variable, but different values for all different stressors)
How should I model this or what statistical test is appropriate for this type of design? In short, I want to check the effect of two discrete variables (plant species and stressor type) on a continuous variable (plant growth), but within one of the two discrete variables (type of stressor), another variable is nested, namely the gradient steps of that stressor. I haven't found any type of ANOVA or lm that tackles this problem.

Thanks in advance!


TS Contributor
You have two initial factors (plant species, stressor type). However, the levels (stressor gradient) are nested within stressor type. Therefore this becomes a third, nested factor. In addition, you have repeated measures (growth over time). You should be able to analyze this as a repeated measures, nested ANOVA or as a repeated measures GLM with a nested factor.