Analysis of binary data - with special attention to Logit model

Abstract

In many studies, the dependent variables are not quantitative, but rather, the response is determined in the sense of the existence or absence of the character under study. Therefore, it is necessary to deal with them and choose the best models of these data that make the advantage in dealing with this phenomenon. Therefore, our current study aimed at finding the best statistical analysis to deal with the two-response data and how to deal with it in terms of evaluation and testing. The study adopted descriptive descriptive method in explaining the models of binary data and clarifying any of these models so that they give results closer to reality.The study found that the probability unit curve, the logistic curve and the exponential curve can be used to represent the probability of response to the study of binary data and to determine the effectiveness of a given variable. The possibilities of the maximum possible means and the reduction of the square of the ki are equal in the large samples, while the preference for the maximum possible means is otherwise. The logit model is generally preferred on the probit model in the software programs for the availability and appropriateness of the software for the first model. As well as logistics conversion is the easiest conversion in the application.