By
Laith. Ali .M. Al-Hayalli
supervised
Dr. Abdul hussian. H . HABEAB . AL-Tai

Abstract
The descriptive dependent variables are important one that do not have quantitative measurement units, they subject to the description of the phenomenon through studying the data and information, and enables the decision–maker to identifying the nature of response when the variable is binary–response or multi–response. The response function may be nonlinear; therefore, there is a need to find other methods to converting it into linear one, this is by means of logarithmic conversion or other transformation that can be converted into linear responses. We used in our thesis the descriptive regression model, when the dependent variable is multi–response, by studying some models related to these descriptive variables. The aim of our thesis is to studying and analyzing the descriptive variable, and its effect on the multi–response dependent variable, by estimation the parameter of model of dependent variable, using weighted least squares method, and the method of maximum likelihood, using “Newton Raphson” and “Jackknife estimation” methods. We used two sides in our study, experimental and empirical, on the experimental we used in our study, experimental and empirical, on the experimental side; we used “The Monte Carlo” method in simulations experiment for three levels of samples (small, middle and large) in different sizes by generate random numbers for the parameter of regression model. In addition, the practical aspect was applied to the life experience of some insecticidal in different concentration, for choosing, then, the best method in reliance on the measurement of Mean Squares Error (MSE), We found, by using The Monte Carlo method of estimation, that the (MLE) method is the best and efficient in the small samples, whereas the (JAK) method is the best in the middle samples, and the two methods of (MLE and (JAK) are equivalent in the large sample size. From the empirical side, on the other hand, we found the result of estimation is that the (MLE) method is the best and efficient because it gets less (MSE).