Experimental Design (Lecture 3)

Question 1

What are the two broad types of experiments?

Answer 1

1. Mensuration “natural” Experiment:
   Opportunistic. Sampling an area where a fire occurred naturally.
2. Manipulation Experiment:
   Taking measurements where a treatment is applied.

Question 2

What is an Experimental Unit?

Answer 2

The smallest unit of the experiment that receives a treatment.

Question 3

What is replication?

Answer 3

Repetition of the experimental unit

Question 4

What is the minimum number of replicates? Why?

Answer 4

Two.

You need at least 1 degree of freedom, therefore, “n” must be equal too, or greater than 2.

Question 5

Why are replications so important?

Answer 5

1. To estimate experimental error/variability (S^2)

2. Insurance against noise

Question 6

What are some examples of noise?

Answer 6

1. Insect outbreak in an experimental unit
2. Plot vandalism
3. Respiratory disease in mice during a drug trial.

Question 7

What are Four key points for when setting up an experimental design?

Answer 7

1. Determine your experimental unit before starting
2. Randomize
3. Get as many replicates as possible
4. Use a control

Question 8

What are the four types of experimental designs?

Answer 8

1. Completely random
2. Randomized block design
3. Latin-Square
4. Systematic

Question 9

What is a completely randomized design?

Answer 9

Treatments are random, but also the experimental units are randomly placed

Question 10

Whats is randomized block design?

Answer 10

1. Experimental units are grouped together in blocks
2. Blocks are stratified or homogeneous
3. As many experimental units as there are treatments

Question 11

Why is the random block design highly recommended?

Answer 11

1. It avoids treatments being lumped together (like in completely randomized)
2. If an entire block is lost, the experiment is not compromized.

Question 12

In randomized block design, what is the preferred orientation of the blocks to an environmental gradient? Why?

Answer 12

Perpendicular. This minimizes the difference between experimental units within the blocks.

Question 13

What is the Latin-Square Design?

Answer 13

1. Used to combat two known gradients
2. All treatments are in each block
3. The same treatment cannot be applied to adjacent blocks.

Question 14

What is a pseudoreplicate?

Answer 14

1. The consequence of assuming that sub-samples are independent and are from independent experimental units.
2. Solution is to have more experimental units.