Deck 12: Logistic Regression

Question 1

What would happen if the age variable was standardised before running the analysis?

A)The overall fit of the model would change
B)The Wald test of significance regarding age would change
C)The odds ratio and log‐odds regarding age would change
D)All of the above would happen

Answer 1

The odds ratio and log‐odds regarding age would change

Question 2

Can continuous variables be included as predictors in logistic regression?

A)No
B)Yes, but only if they are dichotomised into two categories
C)Yes but only if they are cut up into multiple categories
D)Yes

Answer 2

Yes

Question 3

Why are there 3 degrees of freedom for the chi‐square test in the Omnibus Tests of Model Coefficients?

A)Because the current model has 3 more parameters than the baseline model
B)Because the baseline model has 3 more parameters than the current model
C)Because there are 3 significant Wald statistics in the Variables in the Equation box
D)Because the Nagelkerke R‐square is less than .33

Answer 3

Because the current model has 3 more parameters than the baseline model

Question 4

What is the ‐2 Log likelihood for the baseline model?

A)17.55
B)246.04
C)263.59
D)228.49

Question 5

What percentage of accident and emergency attenders are identified by the model?

A)87.3
B)66.5
C)23
D)31.1

Question 6

The next questions refer to the following output that comes from a fictional study predicting accident and emergency attendance over the last 5 years from a scale measuring propensity to take risks, age measured in years) and handedness where right‐handed is the reference category).

Does the model including predictors improve fit over the baseline model?

A)Yes, because the chi‐square test on the Omnibus Tests of Model Coefficients is significant
B)Yes, because the ‐2 log‐likelihood of the model is positive
C)Yes because the overall proportion of correct classifications is bigger than .5
D)No

Question 7

Logistic regression models a linear relationship between a continuous predictor variable and

A)the probability that a dichotomous outcome takes on a particular value
B)the odds that a dichotomous outcome takes on a particular value
C)the log‐odds that a dichotomous outcome takes on a particular value
D)logistic regression cannot include continuous predictor variables

Question 8

Could Body‐mass index BMI) be used as the outcome variable in logistic regression?

A)Yes, if it was dichotomised on a particular cut‐point
B)Yes, as a continuous variable
C)No, because BMI is an ordinal scale
D)No, because BMI may only be used as a predictor variable

Question 9

What can we conclude about the relationship between handedness and accident and emergency visits?

A)Left‐handed people are significantly more likely to attend accident and emergency
B)Right‐handed people are significantly more likely to attend accident and emergency
C)There is no significant relationship between handedness and accident and emergency attendance
D)None of the above

Question 10

In logistic regression, Wald statistics test

A)the overall fit of the model
B)whether an individual predictor makes a significant independent contribution to the model
C)the fit of the baseline model
D)whether the dependent variable is suitable for logistic regression