Question

Using Minitab we obtained the following values, where the values in HI column are leverage values. Using the leverage value method, how many influential observations are there?

           x                      y                  HI      

22                   12                 .38

24                   21                 .28

26                   31                 .22

28                   35                 .20

40                   70                 .92

The stepwise regression procedure

a)   starts with a model with no predictors

b)   starts with a model with all of the predictors c)   starts with a model with one predictor

d)   None of these alternatives is correct

Answer 1

Only 1 influential observation is there.

As it can be seen from the leverage values that only 1 leverage value is very high from rest of the leverage values, we can easily say that only the last point is an influential observation.

2) Starts with a model with 1 predictor

In forward selection we start with a null model and go on adding the terms in it and in backward elemination we start with a full model and go on removing the variables in it until we get the desired model but in stepwise regression we can add or remove the variables depending on their importance.