Some Investigations on Experimental Designs Incorporating Neighbour Effects
KrishiKosh
View Archive InfoField | Value | |
Title |
Some Investigations on Experimental Designs Incorporating Neighbour Effects
Ph D |
|
Creator |
ELDHO VARGHESE
|
|
Contributor |
Seema Jaggi
|
|
Subject |
crop residues, livestock, rice, biological phenomena, methane, inorganic compounds, fractionation, productivity, nitrogen, research methods
|
|
Description |
T-8394
In many agricultural field experiments, neighbour effects from adjacent units is a serious source of bias in making comparison of treatment effects. The way to handle such bias is to develop suitable statistical methodology and designs based on the experimental situations. In response surface methodology, it is generally assumed that the observations are independent and there is no effect of neighbouring units. But in field experiments, the neighbour effects from the treatments applied to adjacent neighbouring units may arise. The procedure for fitting a complete second order response surface model incorporating equal neighbour effects from immediate left and right neighbouring units has been derived. Conditions have been obtained for the estimation of the coefficients of the model. First and second order response surface model have also been studied under the assumption of differential neighbor effects from both left and right units. A method of obtaining designs satisfying the derived conditions has been developed. Blocking aspect in response surface designs in the presence of neighbour effects has been studied and the conditions for orthogonal estimation of parameters of first order model have been derived. Numerical examples have been given for illustrating the experimental situation where experimental units are grouped into complete/ incomplete blocks. There may arise situations where neighbour effect is proportional to the direct effect of treatments. Information matrices have been derived for one-sided and two-sided neighbour effects block model under this set up. The existing neighbour balanced block designs have been studied for different values of the proportionality parameter and their efficiencies have been worked out. The situation where heterogeneity is present in the experimental material in two directions, row-column designs are used. The information matrices for estimating direct and neighbour effects of treatments under different row-column model for directional and non-directional neighbour effects have been derived. Methods of constructing these designs have been developed and their characterization properties have also been studied. These designs are found to be totally balanced/ variance balanced for estimating the direct and neighbour effects of treatments. |
|
Date |
2016-11-01T10:07:03Z
2016-11-01T10:07:03Z 2011 |
|
Type |
Thesis
|
|
Identifier |
http://krishikosh.egranth.ac.in/handle/1/83077
|
|
Format |
application/pdf
|
|
Publisher |
IARI, Some Investigations on Experimental Designs Incorporating Neighbour Effects
|
|