Deck 12: Logistic Regression

Question 1

Logistic regression is most similar to which of the following statistical procedures?

A) Analysis of covariance
B) Canonical correlation
C) MANOVA
D) Discriminant analysis

Answer 1

D

Question 2

Which estimation method is used to estimate parameters in logistic regression?

A) Least-squares estimation
B) Ordinary least-squares estimation
C) Maximum likelihood estimation
D) Likelihood ratio estimation

Answer 2

C

Question 3

What is the technical name of the dependent variable being predicted in logistic regression?

A) A logit
B) An odds ratio
C) A likelihood index
D) A predicted probability

Answer 3

A

Question 4

Which of the following is used to classify cases, based on parameter estimates in a logistic regression model?

A) Predicted probabilities
B) Odds ratios
C) Residuals
D) b coefficients

Question 5

Which of the following would be a preferred coding scheme for a dependent variable in binary logistic regression?

A) Had a fall = 0; did not have a fall = 1
B) Had a fall = 1; did not have a fall = 0
C) Did not have a fall = 0; had 1 fall = 1; had 2+ falls = 2
D) Had a fall = 1; did not have a fall = 2

Question 6

A predictor variable in a logistic regression equation was a birthweight category variable (very low birthweight [1], low birthweight [2], and normal birthweight [3]). With standard (default) contrast coding, how many new variables would be created and what would the codes for normal birthweight be on those variables?

A) Three new variables, with codes of 0, 0, and 1, respectively, for normal birthweight
B) Three new variables, with codes of 0, 0, and 0, respectively, for normal birthweight
C) Two new variables, with codes of 0 and 0, respectively, for normal birthweight
D) Two new variables, with codes of 0, and -1, respectively, for normal birthweight

Question 7

In the scenario described in the preceding question, what would the b coefficient for very low birthweight represent?

A) The effect of very low birthweight compared to low birthweight
B) The effect of very low birthweight compared to normal birthweight
C) The effect of very low birthweight compared to the other two categories, averaged
D) The effect of very low birthweight compared to the average of all categories

Question 8

The null model in logistic regression:

A) Predicts the odds of the outcome based on the entire set of predictors
B) States the null hypothesis that all predictors are unrelated to the outcome
C) Is used as the "perfect" model in the Hosmer-Lemeshow test
D) Predicts the odds of the outcome based solely on the distribution of the outcome in the sample

Question 9

The index -2LL:

A) Is a probability that can range from .00 to 1.00
B) Is the transformed value of the log odds
C) Is used as a basis for assessing the fit of the overall model, in the likelihood ratio test
D) Increases in value as the goodness-of-fit of the model increases

Question 10

Which of the following tests involves deciles of risk?

A) The likelihood ratio test
B) The Hosmer-Lemeshow test
C) The Wald test
D) The logit step test

Question 11

In which of the following tests would the researcher hope for nonsignificance?

A) The likelihood ratio test
B) The Hosmer-Lemeshow test
C) The Wald test
D) The logit step test

Question 12

Which test is most often used to evaluate the significance of individual predictors in logistic regression?

A) The likelihood ratio test
B) The Hosmer-Lemeshow test
C) The Wald test
D) The logit step test

Question 13

Which test is used to evaluate model improvement for added predictors in hierarchical logistic regression?

A) The likelihood ratio test
B) The Wald test
C) The logit step test
D) The standardized residuals test

Question 14

Who developed a formula for calculating the most often reported pseudo R² index?

A) Wald
B) Hosmer and Lemeshow
C) Cox and Snell
D) Nagelkerke

Question 15

Which of the following is an assumption underlying statistical inference in logistic regression?

A) Multivariate normality
B) Homogeneity of variance
C) Linearity between predictors and the dependent variable
D) Linearity between predictors and the logit

Question 16

Inspection of standardized residuals would be used to evaluate which of the following?

A) Assumption of linearity
B) The presence of outliers
C) Possible multicollinearity
D) The cut point to use to classify cases

Question 17

Questions pertain to the following table (Table 12), which presents results for a fictitious logistic regression analysis predicting acute confusion in long-term care residents.

-Could predicted probabilities for each case be computed based on information in Table 12?

A) Yes, all necessary information is provided.
B) No, because the Wald statistic is not provided.
C) No, because the likelihood ratio test chi-square is not provided.
D) No, because the constant is not provided.

Question 18

Questions pertain to the following table (Table 12), which presents results for a fictitious logistic regression analysis predicting acute confusion in long-term care residents.

-Refer to Table 12. Which of the following variables was not a significant predictor of acute confusion?

A) Inadequate fluid intake
B) Having had a fall in the past month
C) Taking five or more medications
D) Vision impairment

Question 19

Questions pertain to the following table (Table 12), which presents results for a fictitious logistic regression analysis predicting acute confusion in long-term care residents.

-Refer to Table 12. If the Wald statistic had been reported, which predictor would be most likely to be significant at p < .001?

A) Inadequate fluid intake
B) Having had a fall in the past month
C) Taking five or more medications
D) Vision impairment

Question 20

Questions pertain to the following table (Table 12), which presents results for a fictitious logistic regression analysis predicting acute confusion in long-term care residents.

-Refer to Table 12. What information in the table indicates that the overall predictive model was a sufficiently good fit to the data?

A) The odds ratios
B) The Hosmer-Lemeshow results
C) The Nagelkerke R² results
D) The classification results