Question

Critically discuss differences between logit and tobit estimators.

Answer 1

• Logit Models

The models used for distinct outcome modeling are known as logit models. Under this modeling, sometimes binary outcomes or otherwise three or more than three outcomes may occur. In binary outcomes, 0 and 1 are the outcomes. All the other outcomes are known as multinomial logit. This logit model works according to logit distribution. It is also known as Gumbel distribution. Under this kind of distribution, sample sizes of large lot are usually preferred.

• Tobit Models

Tobit model is entirely different from logit model. This model is also known as censored regression models. This model helps to find out linear relationships among variabes accompanied with left or right censoring occuring in the dependent variable. Under this model, there will not be any binary outcomes or other distinct outcomes. These are a kind of linear regression. Hence this model will be used only when a continuous dependent variable have to be regressed which has to be change the position of the direction. This model allows such a variable regression during censoring and thereby helps in the regression of continous dependent variable. It helps the analyst of this model to determine the threshold,whether lower or upper,for censoring this regression along with keeping a balance on linear assumptions which are required for such a linear regression.