Please help me to briefly explain the difference between "predicting the mean response" and "predicting the response of a new case." If confidence intervals are computed for both type of predictions at the same x-values at the same confidence level, which interval is wider? Why?