AGRC1023 Lecture Notes - Lecture 6: Nuisance Variable, Anthology Series, Statistical Unit
Document Summary
If looking at 3 varieties and 4 rates of treatment, it"s 12 treatment combinations. We replicate experiments to ensure results are consistent. Number of replicates depends on original study sample size, dictated by cost, resources and ability to collect data in reasonable time. More than one measurement is made within experimental unit and not accounted for in design. Well-defined protocols (time, temperatures) under which study is conducted. Noting pests, other influencers, that may affect results o. Removes some experimental error by ensuring experimental units are blocked/placed closely, they are more likely to experience the same environment so results are more likely to be reliable but randomise within the block block. Balanced design has full set of treatments within single block. Randomised block design only one primary factor under consideration in experiment. Blocking nuisance factors enables removing nuisance variation from experimental units. Block design maintains internal validity, reducing possibility that observed effects due to confounding source of variation.
