Comparison effect of error whit Cauchy and normal distribution in linear regression models

One of the disputable issues in the analysis of linear models is that the error distribution is not normal. Because the normal distribution has thin-tailed, this model is sensitive to outlier data and issues affects the inferred parameters. For this reason, in recent years other robust methods have been developed. One of the ways is the use the heavy-tail distributions as Cauchy exist normal distribution. Replacing the Cauchy distribution instead of error normal distribution in linear model reduce the effect of outlier data. Hence use the heavy-tail distribution in inference the parameters is more appropriate. According to the results show the model with normal error is better when we dont have outlier data but when, exist outlier data the model with Cauchy error has DIC less than the model with normal error.