Augmented Designs-Experimental Designs in Which All Treatments are not Replicated
Burgueño, J.
Augmented Designs-Experimental Designs in Which All Treatments are not Replicated - Madison, WI. (USA) : American Society of Agronomy; Crop Science Society of America; Soil Science Society of America, Inc., 2018.
In the research planning stages of a project, sometimes replication of treatments is impractical, prohibitively expensive, or impossible. Perhaps the researcher does not have enough material to fully randomize and replicate, or there are physical space and/or time limitations, or too many barriers to complete the data collection required for a fully replicated experiment to answer the proposed research question. In these scenarios, it can be advantageous to use unreplicated or augmented experimental designs. These designs reduce costs with an acceptable loss of precision. Other (though not all) important consequences are: (i) not all treatments are compared with the same precision; (ii) experimental error is partially estimated; (iii) in factorial experiments, some interaction effects cannot be estimated and some are confounded; (iv) standard experimental designs can (and should) be modified to deal with unreplicated treatments; and v) the statistical analysis approach is usually different to the usual analysis of variance.
Text in English
978-0-89118-360-0
https://doi.org/10.2134/appliedstatistics.2016.0005
Experimental design
Statistical methods
Experimentation
Augmented Designs-Experimental Designs in Which All Treatments are not Replicated - Madison, WI. (USA) : American Society of Agronomy; Crop Science Society of America; Soil Science Society of America, Inc., 2018.
In the research planning stages of a project, sometimes replication of treatments is impractical, prohibitively expensive, or impossible. Perhaps the researcher does not have enough material to fully randomize and replicate, or there are physical space and/or time limitations, or too many barriers to complete the data collection required for a fully replicated experiment to answer the proposed research question. In these scenarios, it can be advantageous to use unreplicated or augmented experimental designs. These designs reduce costs with an acceptable loss of precision. Other (though not all) important consequences are: (i) not all treatments are compared with the same precision; (ii) experimental error is partially estimated; (iii) in factorial experiments, some interaction effects cannot be estimated and some are confounded; (iv) standard experimental designs can (and should) be modified to deal with unreplicated treatments; and v) the statistical analysis approach is usually different to the usual analysis of variance.
Text in English
978-0-89118-360-0
https://doi.org/10.2134/appliedstatistics.2016.0005
Experimental design
Statistical methods
Experimentation