| In this thesis the study is extended around the feeding-and-management condition and indexes. It started with static analysis of the condition of feeding and management 1. The optimized feeding-and-management condition was selected only allowing for production performance and the changes of feed consumption. 2. The selection of optimized feeding-and-management condition and major effect analysis under multiple indexes. And the next is dynamic analysis of the optimized feeding-and-management condition. 1. The feeding-and-management condition is quantum factor and the index is static. 2. The feeding-and-management condition is qualitative factor and the index is dynamic. 3. The feeding-and-management condition is quantum factor and the index is dynamic. We try to deepen our acquaintance about the utilization of me statistical method on the selection of optimized feeding-and-management condition through this study and to give a theoretical evidence for the design, test index and test method of animal nutrition experimental.The results showed as follows:1. The selection only allowing for production performance and the changes of feed consumption.Take economical effect as index to adjust the feeding -and-management condition. Assume that the ratio of animal production and the feed price is k ? 1, the both price described as ok and a . Then the profit p=aky-ax. And taking the profit as index to conduct analysis of variance we can get a value interval of k when mere are no significant difference between each two conditions of feeding and management byIf A < 0 . k take any value above d, the difference between the two conditions is significant. If A < 0 , letWhen d e (k], k2), the value interval of k that can ensure the difference of the two conditions null significant is (d, k2), when d e (-cc, k}), the value interval of k that ensure the difference between two conditions significant is (&,.&,), and whend e (k2 ,-H?, then when k take a value above d, the two conditions have significantdifference.Where, the ratio of feed to meal of this breed is d ? 1.2. The selection of optimized feeding-and-management condition and major effect analysis under multiple indexes.By using multiple statistics and comparison the total difference of different levels or different treats was inferred. But for getting more concrete statistical result, it also should be known if the indexes have significant difference between different levels or treats. Since there always has certain correlations between the indexes which reflecting the same treat effect, it must influence the comparability of the index when compare the difference of a index in different level or treat if other indexes has difference in different levels or treats of. So unitary covariance analysis should be adopted to compare the difference of each index in different level or treat.3. The feeding-and-management condition is quantum factor and the index is static.As to the experiment of quantum factor, regression model can reduce the possibility of committing type I error because it take the correlation of factors and indexes into account compared with analysis of variance. It is showed by the reduction in the free degree of the treatmean square of FA . And also because of its function of forecast, which is not in analysis ofvariance, regression analysis should be used to treat the experiment of quantum factor.4. The feeding-and-management condition is qualitative factor and the index is dynamic.When taking a value to the change of unlimited dynamic index, establishing a mathematic model, the multiple statistic analysis can be used to compare the difference of model parameters in different treat simultaneously. So the comparison process of dynamic index can be simplified, and the effect of statistic test can also be improved because the correlation of each parameter is taken into account5. The feeding-and-management condition is quantum factor and the index is dynamic. 5.1 The selection when change the price of production... |
PDF Full Text Request