Regression Analysis
Many models are being set up by the researchers in order to analyze the relationship between a response and some predictors. When the response variable is continuous linear regression is used. Remaining errors follow the normal distribution is one of the assumptions of linear regression and this assumption will fail if the response variable is categorical therefore this is the reason that an ordinary linear model is not used. In an ordinary linear regression, there is the linear relationship between the coefficient and the response variable that will represents the predictor variable.
In case of dichotomous response variable, similar linear model is set up if numerical values are used to represents the two categories. For ease the values of 0 and 1 are chosen as we would give y=1 then it means that the plant would lives and if we would give y=0 then it means that plant dies. This linear model does not work well and face different problems for many reasons for example as we find that the plant with high level of fungal infection will fall into the category “plant live” more often than those plant with low level of infection. Thus this will increase the level of rise in infection and thereby the probability of plant living decreases.
However the general increase in the infection level is accompanied by the general decrease in the probability and also we know that all the probability will fall under 0 and 1. Therefore it is important to assume the relationship between p and x1 to be sigmoid instead o straight line. It will help you to determine the relationship between the x1 and the function of p. This article contains the regression analysis and it will help you to analyze the regression in much easier way. This article contains certain assumption of linear regression which gives different problems to the model.
Written by: Matt
We also suggest this relevant article if you have time: Expectancy Theory by Victor Vroom
Some other similar articles
Tagged as assumption, assumptions, fungal infection, linear model, linear regression, linear relationship, models, normal distribution, numerical values, probability, regression analysis, straight line, x1 + Categorized as Economy articles