parts of an experiment
key elements of a good design
treatment
variable being tested
control
baseline for comparision
replication
repeating treatments to confirm results
randomization
assigning treatments randomly to avoid bias
block design types
completely randomized design
treatments assigned randomly. used in uniform conditions
randomized complete block design
blocks are perpendicular to the gradient, each block contains all the treatments randomized within it. ideal for fields with one gradient
latin square design
treatments arranged so each appears once per column and row. used when 2 gradients exist
pseudoreplication
where multiple measurements within the same plot are incorrectly treated as independent replicates