الفهرس | Only 14 pages are availabe for public view |
Abstract SUMMARY AND CONCLUSION Relative precision, i.e., relative efficiency of different experimental designs has been investigated by many researchers. In Egypt results from uniformity trials with different field crops have been reported by several investigators, such as, El-Kadi. (1973). El-Rassas (1982) and Moursi et ale e 1983) . In cotton, several trials have also been ~onducted. However, some major points have not as yet being calculated such as the efficiency of covariance ~nalysis, regression analysis arid guarded hills procedure.; Thus, thiS work has been conducted with the objective of evaluating the relative efficiency of different experimental. designs and techniques in experimental field work of cotton . . ..Two.~fleld”expe[’imental tr.ials. were_conducted. a uniformity trial in 1982 and the second was designed in randomized complete block design with six replications in 1983. In the first trial of 1982. Giza 75 the cotton cultivar was used. The second trial of 1983 included in addition to Giza 75. the cotton cultivars Giza 69, Giza 70 and Giza 45, Data collected included seedcotton yield and some major components of yield. These were utilized in computing relative efficiency estimates. Results are summarized as follows: 5.1. The relative efficiency of experi~ental designs: The randomized complete block design was more efficient than the completely randomized design (106.47%). - The latin square design was more efficient than the randomized complete block (110.45%). Also, the relative efficiency of rows was higher than that of columns. - The split-plot, split-split-p1ot and strip plot designs were of higher efficiency as to their sensitivity to measure interaction (115.38, 131.79 and 143.40% for splitplot, split-split-plot and strip-p16t designs, respectively). - The incomplete block designs were more efficient than the .’ randomized complete block design (126.20%). - The balanced incomplete blocks were more efficient than the partially balanced designs (133.88% and 141.91% for balanced and partially balanced, respectively). - The double control procedure warranted a higher degree of efficiency than single control (123.97% and 113.25% for double control and single control, respectively). - The” square- lattices expressed h~igher-e’fficiency than-the rectangular lattices (132.22% and 97.60% for square lattices and rectangular lattices, respectively). - The Lat t tce square designs were more efficient than the rectangular lattices. 5.2. The relative efficiency of plot and block size and shape: - The rectangular large plots were of higher efficiency (140.34%), especially when the length of plots runs in parallel to the direction of soil heterogeneity in experimental units. An inverse relationships between both plot sizes and number of treatments and coefficient of variability was observed. - There is no clear difference in efficiency of square blocks or rectangular blocks. 5.3. The relative efficiency of covariance analysis and guarded hills procedures: _ Covariance analysis (ANCDVA) was found more erficient than the analysis of variance (ANOVA) for adjusted treatment means. - The guarded plants procedure was more efficient’than its counterpart of unguarded plants. 5.4. The relative efficiency of multiple linear regression: - Model II was found more efficient than Model I. This may be attributed to the fact that the number of bolls per branch is more essential in determining seedcotton yield than the length of fruiting branch per plant. - Other components were of similar weight in both models. - All genotypes tested gave the same trend in both seasons by using the two models. This indicates that the importance of yield components variables fitted in the model also, do not change in relation to seasonsl variations. On the other hand, there is no effect of the seasonal variations on different studied characters of yield components. In comment, the findings reported herein coincide with the moderntheo ry of expe rimen tal design that calls for randomization, local control and central of experimental error with a minor deviation, i.e. results on randomization vs. systematic distribution of treatments wereir~es~lute. This, however will not underestimate ~he importance of randomization in experimental design because randOUiization is necessary to destroy the correlation among errors and make valid the usual tests of significance. Also, the importance of block shape on precision was not ·clear. Various uniformity trial studies, with different field crops have shown that the block whether complete or . incomplete should be apprOXimately square. The findings of this work agree with thos~ previously reported by others. In addition to that, the importance of guarded hills procedure was made clear. No advantage of guarded hills over unguarded hills procedure was observed. The reason given was that ~he relationship of number of plants on a given area and seedcotton yield per plant is compensatory in its nature. However, data reported herein showed undoubtdly the importance of guarded hills procedure. |