The following points highlight the nine criteria used for comparison of various experimental designs. The criteria are: 1. Condition of Use 2. Layout 3. Replications 4. Randomization 5. Local Control 6. Statistical Analysis 7. Components of Variation Estimated 8. Accuracy of Results Obtained 9. Efficiency.
Criteria # 1. Condition of Use:
The use of experimental design depends mainly on the amount of experimental material to be tested and nature of fertility variation of the field. The adoption of CRD is ideal when the experimental material is limited and homogeneous.
RBD can be used when the fertility variation moves in one direction and adoption of LSD is useful when the fertility variation moves in two directions. The SPD is used when several factors are to be tested simultaneously, out of which some require larger plots and others require smaller plots.
Criteria # 2. Layout:
The plan of layout differs from design to design. In case of CRD the whole field is divided directly into plots of equal size. The number of plots is equal to the product of replications and treatments. In case of RBD first the field is divided into homogeneous blocks equal to the number of replications.
Then each block is divided into plots equal to the number of treatments. In LSD the field is divided into equal number of rows and columns. The number of rows and columns is equal to the number of treatments. For the layout of SPD the field is first divided into homogeneous blocks equal to the number of replications.
Then each block is divided into plots equal to the number of first factor. Then each main plot is divided into sub-plots equal to the level of second factor, and then each sub-plot is divided into ultimate plots equal to the level of third factor.
Criteria # 3. Replications:
The number of replications to be used also differs from design to design. In case of CRD, the number of replications may differ from treatment to treatment. However, this flexibility in the choice of replications is not possible for other designs. In case of LSD the number of replications has to be equal to the number of treatments. Generally 4 or 5 replications are sufficient in case of CRD, RBD and SPD and 5-12 in case of LSD.
Criteria # 4. Randomization:
In case of CRD randomization is done treatment-wise. In RBD randomization is done replication or block-wise. Separate randomization is used in each block. In case of LSD, randomization is done with help of reduced latin square and then rows, columns and treatments are reshuffled with the help of random numbers.
In case of SPD, the levels of first factor are randomized block wise. The levels of second factor are randomized main plot wise and the levels of third factor are randomized subplot wise.
Criteria # 5. Local Control:
The principle of local control is not adopted in case of CRD. The error variance can only be controlled by selecting a homogeneous set of experimental units. Selection of homogeneous units is not possible when number of treatments is large. In other designs like RBD. LSD and SPD, the principle of local control is adopted by forming homogeneous blocks.
Criteria # 6. Analysis:
The analysis or calculation is easy in case of CRD. The analysis remains simple even when the results of some units are missing. In case of RBD, the analysis is easy but it becomes complicated when missing plot technique is applied. In case of LSD, the analysis is also simple but it becomes complicated when several plot yields are missing. In SPD, the analysis is complicated as compared to other designs.
Criteria # 7. Components of Variation:
In case of CRD, total variation is divided into two components, i.e., treatment and error. In RBD, the total variation is divided into three components, viz., blocks, treatments and error, while in case of LSD the total variation is divided into four components, viz., rows, columns, treatments and error. In case of SPD, total variation is split into several components.
Criteria # 8. Accuracy:
CRD is suitable for pot culture experiments only. RBD provides more accurate results than CRD due to formation of homogeneous blocks and separate randomization in each block. LSD provides highly precised results because the fertility variation is controlled in two directions which reduce the standard error. SPD provides different precisions for different factors.
Criteria # 9. Efficiency:
The efficiency of different designs varies depending upon the number of treatments. In CRD a large number of treatments can be tested if the experimental material is other than field. In case of RBD the efficiency of error control decreases beyond 20 treatments due to increase in heterogeneity within the block.
The maximum number of treatments for which we can use LSD is 12. In case of SPD several factors can be tested with increasing trends of precision.