Deck 11: Logistic Regression

Question 1

Logistic regression is basically an extension of multiple regression in situations where the DV is not a continuous or quantitative variable.

Answer 1

True

Question 2

Logistic regression is also sometimes used as an alternative to discriminant analysis.

Answer 2

True

Question 3

The value that is being predicted in logistic regression is a probability, which ranges from 0 to 1.

Answer 3

True

Question 4

Like both discriminant analysis and multiple regression, logistic regression requires that assumptions about the distributions of the predictor variables need to be made by the researcher.

Question 5

Logistic regression is unable to produce nonlinear models.

Question 6

The chi-square goodness-of-fit test compares the actual values for cases on the DV with the predicted values on the DV.

Question 7

Smaller values on the -2 Log Likelihood indicate that the model fits the data better.

Question 8

Cox & Snell R Square and Nagelkerke R Square are essentially estimates of R² indicating the proportion of variability in the DV that may be accounted for by all predictor variables included in the equation.

Question 9

The classification table compares the predicted values for the IVs, based on the logistic regression model, with the actual observed values from the data.

Question 10

The significance of each predictor is tested with a t test as in multiple regression.

Question 11

The odds ratio represents the increase (or decrease if Exp(B) is less than 1) in odds of being classified in a category when the predictor variable increases by 1.

Question 12

Even though logistic regression does not require the adherence to any assumptions about the distribution of predictor variables, several problems may occur if too few cases relative to the number of predictor variables exist in the data.

Question 13

Logistic regression may produce extremely large parameter estimates and standard errors, especially in situations where combinations of discrete variables result in too many cells with no cases.

Question 14

Logistic regression is not sensitive to high correlations among predictor variables.

Question 15

Logistic regression is also not sensitive to outliers.

Question 16

Probabilities are simply the number of outcomes of a specific type expressed as a proportion of the total number of possible outcomes.

Question 17

In a logistic regression application, odds are defined as the ratio of the probability that an event will occur divided by the probability that the event will not occur.

Question 18

Probabilities will always have values that range from 0 to 1, but odds may be greater than 1.

Question 19

The ultimate model obtained by a logistic regression analysis is a linear function.

Question 20

Wald is a measure of association for B and represents the significance of a variable in its ability to contribute to the model.

Question 21

The results summary should always describe how variables have been transformed or deleted.

Question 22

The accuracy of classification should also be reported in the narrative.

Question 23

The results section in logistic regression should contain a table that includes B, Wald, df, level of significance, and odds ratio.

Question 24

In binary logistic regression, the DV may be dichotomous and the IVs may be continuous or categorical.

Question 25

A good-fitting model in logistic regression will typically have fairly low values for -2 Log Likelihood, significant model chi-square, and variables with odds ratios greater than or equal to 1.

Question 26

Which of the following responses is not true of logistic regression?

A) It is basically an extension of multiple regression in situations where the DV is not a continuous or quantitative variable.
B) Like both multiple regression and discriminant analysis, logistic regression requires the same assumptions about the distributions of the predictor variables (IVs) to be made by the researcher.
C) This procedure is also sometimes used as an alternative to discriminant analysis.
D) The basic concepts that are fundamental to multiple regression analysis-namely that several variables are regressed onto another variable using one of several selection processes-are the same for logistic regression analysis, although the meaning of the resultant regression equation is considerably different.

Question 27

The main output components to interpret in the results obtained from a logistic regression analysis include which of the following?

A) The resulting model, using goodness-of-fit tests.
B) A classification table for the DV.
C) The table of coefficients for variables included in the model.
D) Responses (a), (b), and (c) are correct.

Question 28

Mathematically speaking, logistic regression is based on:

A) Probabilities.
B) Odds.
C) The logarithm of the odds.
D) All three above responses are correct.

Question 29

Which of the following statements is incorrect?

A) Probabilities are simply the number of outcomes of a specific type expressed as a proportion of the total number of possible outcomes.
B) Odds are defined as the ratio of the probability that an event will occur divided by the probability that the event will not occur.
C) Odds will always have values that range from 0 to 1.
D) The odds ratio is defined as a ratio of the odds of being classified in one category of the DV for two different values of the IV.

Question 30

Which of the following is true in the interpretation of results?

A) Cox & Snell R Square and Nagelkerke R Square represent two different estimates of the amount of variance in the DV accounted for by the overall model.
B) Chi-square statistics with levels of significance are computed for the model, block, and step.
C) Wald is a measure of significance for B and represents the significance of each variable in its ability to contribute to the model.
D) All of the above are true.