Components/Classification of Experimental Design
In an experiment, the examination of responses to determined treatments and environmental conditions can face several obstacles. These obstacles include the natural variability inherent in each object and the influence of various external factors that cannot be arranged for all objects in the experiment. In this case, statistics can be a tool for researchers to separate and investigate sources of response variability, including parts caused by treatment, environment, and other influences that cannot be clearly identified.
There are three key components to consider in experimental design:
- Treatment Design: Specific conditions that are deliberately created to induce a response. Treatment can be interpreted as specific conditions given to experimental units. This relates to how the treatment is structured, such as through single factor, factorial, split plot, or split block. The treatments are generally designed in a crossed or nested structure.
- Treatments are designed in a crossed (crossed) structure or factorial pattern if each level of one treatment appears at each level of another treatment. For example: If Treatment A has 6 levels, and Treatment B has 3 levels, then the cross-treatment design is as follows:
B A 1 2 3 4 5 6 1 x x x x x x 2 x x x x x x 3 x x x x x x Or in a horizontal form::
A
1
2
3
4
5
6
B
B
B
B
B
B
123
123
123
123
123
123
xxx
xxx
xxx
xxx
xxx
xxx
If Treatment A and Treatment B are also crossed with Treatment C (for example: 2 levels):
C
1
2
A
A
1
2
3
4
5
6
1
2
3
4
5
6
B
B
B
B
B
B
B
B
B
B
B
B
123
123
123
123
123
123
123
123
123
123
123
123
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
Treatment B is nested within Treatment A if different levels of Treatment B appear only once within a level of Treatment A, for example:
A
1
2
3
4
B
B
B
B
1
2
3
4
5
6
7
8
9
10
11
12
x
x
x
x
x
x
x
x
x
x
x
x
Treatment B, consisting of 12 levels, is nested within 4 levels of Treatment A. In this nested structure, it's possible for the design to be unbalanced, for instance at level 3 of Treatment A there are only 2 levels of Treatment B, while the others have 3 levels of Treatment B.
A
1
2
3
4
B
B
B
B
1
2
3
4
5
6
7
8
9
10
11
x
x
x
x
x
x
x
x
x
x
x
Nested patterns do not have interactions!
- Experimental/Environmental Design: Environmental conditions and the natural variability of objects that can obscure or disrupt the examination of the responses that emerge. This design relates to how treatments are placed on experimental units (e.g., Completely Randomized Design (CRD), Randomized Block Design (RBD), Latin Square Design (LSD), Lattice).
- Response Design: The response given by the experimental object. This relates to determining the characteristics of experimental units that will be used to assess or measure the influence of treatments.
Things to know in Response Design:
- Must reflect the influence being studied. Suppose you are conducting an experiment about the effect of cow manure fertilization on corn growth. Then you should create a response design that can reflect the influence of this manure on corn growth. For instance, plant height, leaf count, leaf area, etc.
- There are measurement scales:
- Qualitative: nominal and ordinal (cannot be analyzed by variance). This scale is subjective and the guidelines for conducting measurements are mostly non-standard.
- Quantitative: interval/range and ratio. This scale is objective and its measurement tools are often available.
- There are observation units, which are the smallest units used in measurement.
- There are evaluation units, which are the smallest units representing experimental units used in data analysis, or evaluation units are the average of observation units.
The type of treatment can be divided based on its nature and quantity. Based on its nature, treatment can be qualitative (e.g., type of fertilizer, variety, method of soil treatment) or quantitative (e.g., fertilizer dosage, volume of pesticide). Based on its quantity, treatment can be a single factor (only one factor being studied) or factorial (consisting of two or more treatments).
In response design, there are things to know such as the selection of characteristics that will assess or measure the influence of treatments, the presence of a measurement scale (qualitative or quantitative), the presence of an observation unit as the smallest unit in measurement, and the presence of an evaluation unit as the smallest unit representing experimental units in data analysis.
By understanding these three components: treatment design, environmental design, and response design, researchers can design and conduct experiments more efficiently and effectively, and minimize potential biases or disturbances that may arise.
