Chapter 19 - Logistic Regression

Question 1

We use logistic regression to predict an outcome variable that is__________ from one or _________ __________ or continuous predictor variables.

Answer 1

We use logistic regression to predict an outcome variable that is CATEGORICAL from one or MORE CATEGORICAL or continuous predictor variables.

Question 2

We cannot have a categorical outcome variable otherwise because it

Answer 2

violates the assumption of linearity in normal regression

Question 3

a logistic regression predicts the..

So a value of 0 means… and 1 means…

Answer 3

probability of an outcome occurring

- 0 means something is very unlikely…and 1 means very likely

Question 4

The log likelihood is an indicator of how much__________ information there is after the…

Answer 4

UNEXPLAINED INFORMATION after the model has been fitted

Question 5

Large log likliehoods indicate

Answer 5

poorly fitting statistical models

Question 6

The R statistic is the _______ _________ between the outcome variable and each of the predictor variables. It can vary between:

Answer 6

• R stat = partial correlation

- between 1 and -1

Question 7

When we subtract the deviance of a new model from the baseline deviance we have the

Answer 7

log-likliehood ratio

Question 8

The Log-likliehood ratio has a degrees of freedom = to…

Answer 8

the number of k in the new model minus the number of k in the old model which is always 1

Question 9

The number of k in the base model is

We use the k of the new and old to get the df for a

Answer 9

• always 1 because it only has the constant in it

- Chi square distribution

Question 10

In Logistic Regression you should NEVER square the

Answer 10

R statistic

Question 11

the deviance has a

Answer 11

chi square distribution

Question 12

striving for parsimony means

Answer 12

predictors should not be included unless they have explanatory benefit