An Evaluation of Control Plots and Moving Means for Error Control in Barley Nurseries¹

Abstract

Frequently repeated control plots and the moving mean of adjacent plots were compared as methods of controlling error in single‐row plots of barley, Hordeum sativum Jess. (H. vulgare L.) Adjustment by either technique was more effective for grain protein content than for grain yield. Covariance adjustment using either moving mean or control plots prevented overadjustment in several instances and was always more effective than percentage adjustment in reducing experimental error. Repeated control plots and moving means gave similar control of experimental error, but control plots had less practical application than moving means. A partially balanced lattice design accomplished superior control to both methods using percentage adjustment and similar control using covariance adjustment. In cases where the number of entries was too large to accommodate an incomplete block design, covariance adjustment using the moving mean would be advantageous.

Fundamentals of Experimental Design: Guidelines for Designing Successful Experiments

“…Researchers organizing these types of field experiments have three options: (i) conduct the experiments on multiple farms, using farms as blocks; (ii) use control-plot designs in which one control treatment is interspersed with the other treatments, optimally in an every-other-plot pattern; or (iii) use a combination of both approaches. Control-plot designs were developed in the early 20th century (Pritchard, 1916;Richey, 1924) and their popularity lasted through the 1970s, largely for use in evaluating extremely large numbers of lines or families for plant breeding (Baker and McKenzie, 1967;Mak et al, 1978). These designs fell out of favor when it became clear that devoting half of the experimental units to a single treatment or cultivar was highly inefficient (Melton and Finckner, 1967), and modern spatial analysis methods were just emerging (Gaeton and Guyon, 2010).…”

“…9, 10, and 11) provided a thorough treatment of these designs, complete with numerous plans and analytical details. Lastly, control-plot designs (Baker and McKenzie, 1967;Mak et al, 1978), augmented designs (Murthy et al, 1994;Scott and Milliken, 1993;Wolfinger et al, 1997), and modified augmented designs Poushinsky, 1983, 1985;Lin and Voldeng, 1989) were designed to accommodate large numbers of treatments in which there is considerable lack of uniformity in the number of replicates per treatment or some treatments are completely unreplicated.…”

Tolerância de cana-de-açúcar a herbicidas avaliada pela diferença dos tratamentos

“…No setor sucroalcooleiro, mesmo tendo o conhecimento de que os cultivares apresentam respostas distintas de tolerância a herbicidas (Oliveira et al, 2004;Ferreira et al, 2005), os programas de melhoramento genético focam a seleção dos novos genótipos principalmente para fatores edafoclimáticos, pragas e doenças (Matsuoka et al, 2005), não considerando a tolerância dos cultivares. A escassez de recursos humanos e de material para propagação, o longo ciclo da cultura e as extensas áreas experimentais são as principais justificativas para a ausência da informação.…”

“…Segundo Mak et al (1978), ao estudarem linhagens de cevada, o método possui precisão semelhante à dos métodos de médias móveis e delineamento em látice. Contudo, a técnica das testemunhas pareadas apresenta o aumento da área experimental como a desvantagem do método.…”

Efficiency of augmented designs for selection

“…Townley- Smith and Hurd (1973) compared the efficiencies of adjusting plot yields via repeated checks and moving averages, and found that the latter method gave the best control of error variance. Mak et al (1978) found that adjustment by either method was superior to no adjustment for grain protein content, but not for grain yield. Both techniques gave similar control of experimental error.…”