Why Randomize Agricultural Experiments

This study illustrates the importance of randomization using two hypothetical field trials, one with a marked systematic trend and the other with a more erratic spatial pattern. The insights from these two examples are reinforced by analysis of a uniformity trial and a small simulation study. Results illustrate that both model-based spatial analysis and randomization-based analysis assuming independent errors are valid with full randomization but may be invalidated when randomization is lacking. It is concluded that randomization provides protection against different forms of spatial trend. The examples given in the study serve as a general reminder that agricultural experiments should be randomized whenever possible.

[1]  Hans-Peter Piepho,et al.  Statistical aspects of on-farm experimentation , 2011 .

[2]  H. Piepho,et al.  Linear variance models for plant breeding trials , 2010 .

[3]  E. Williams,et al.  The use of uniformity data in the design and analysis of cotton and barley variety trials , 1988 .

[4]  David R. Cox,et al.  Randomization in the Design of Experiments , 2009 .

[5]  Are neighbour methods preferable to analysis of variance for completely systematic designs? ‘Silly designs are silly!’ , 1995 .

[6]  H. F. Smith.  An empirical law describing heterogeneity in the yields of agricultural crops , 1938, The Journal of Agricultural Science.

[7]  George A. Milliken,et al.  On the analysis of long‐term experiments , 2007 .

[8]  Brian R. Cullis,et al.  Accounting for natural and extraneous variation in the analysis of field experiments , 1997 .

[9]  Hans-Peter Piepho,et al.  A Mixed Modelling Approach for Randomized Experiments with Repeated Measures , 2004 .

[10]  Hans-Peter Piepho,et al.  Getting the Most Out of Sorghum Low-Input Field Trials in West Africa Using Spatial Adjustment , 2012 .

[11]  H. Piepho,et al.  Visual Scorings of Drought Stress Intensity as Covariates for Improved Variety Trial Analysis , 2013 .

[12]  Michael G. Kenward,et al.  An improved approximation to the precision of fixed effects from restricted maximum likelihood , 2009, Comput. Stat. Data Anal..

[13]  R. N. Edmondson,et al.  Past developments and future opportunities in the design and analysis of crop experiments , 2005, The Journal of Agricultural Science.

[14]  R. A. Bailey,et al.  VALID RANDOMISATION FOR THE FIRST‐DIFFERENCE ANALYSIS , 1996 .

[15]  NEIGHBOUR ANALYSIS OF UNIFORMITY DATA , 1986 .

[16]  Geostatistical Models in Agricultural Field Experiments: Investigations Based on Uniformity Trials , 2012 .

[17]  E. Pitman.  SIGNIFICANCE TESTS WHICH MAY BE APPLIED TO SAMPLES FROM ANY POPULATIONS III. THE ANALYSIS OF VARIANCE TEST , 1938 .

[18]  J. John,et al.  Construction of Resolvable Spatial Row–Column Designs , 2006, Biometrics.

[19]  Chris Brien,et al.  Formulating mixed models for experiments, including longitudinal experiments , 2009 .

[20]  H. Piepho,et al.  The Effect of Preceding Crop and Pre‐Sprouting on Crop Growth, N Use and Tuber Yield of Maincrop Potatoes for Processing Under Conditions of N Stress , 2007 .