What is true about studentized residuals in multiple linear regression?

In multiple linear regression, studentized residuals are the residuals of the model divided by an estimate of their standard deviation. This transformation typically makes the distribution of these residuals approximate a t-distribution, which is particularly important for hypothesis testing and identifying outliers.

Using a t-distribution allows for the consideration of the degrees of freedom, which accounts for the sample size and the number of parameters estimated in the model. Because studentized residuals are standardized, they facilitate comparison across cases within the dataset, making it easier to detect outliers and assess the fit of the regression model.

The other options are not applicable to studentized residuals. For instance, these residuals do not follow a binomial distribution, which is associated with binary outcome variables. Additionally, they are not measured in the same unit as fitted values; instead, they are dimensionless because they are standardized. Finally, while a negative residual indicates that the observed value is below the predicted value, it does not necessarily imply that there's a probable outlier; outlier detection requires further investigation beyond simply looking at the sign of the residual.