Question 1

Which of the following would be an appropriate outcome to examine with binary logistic regression?&#10;A) homeownership (own, rent)&#10;B) location of exercise (gym, home, outside)&#10;C) preferred art (dance, studio art, theatre, other)&#10;D) vacation location (beach, mountains, theme park, other)

Accepted Answer

Variables with only two categories can be used with binary logistic regression.

Question 2

A researcher is studying passing rates for a licensure exam and has coded 'pass' as '1' and 'fail' as '0'. Which of the following is a correct interpretation of odds of .48?&#10;A) The odds of passing are about equal to those of failing&#10;B) The odds of passing are greater than those of failing&#10;C) The odds of passing are less than those of failing&#10;D) Cannot be determined from the information provided

Accepted Answer

The odds of passing are about those of failing (with .50 being exactly equal).

Question 3

A researcher is studying passing rates for a licensure exam and has coded 'pass' as '1' and 'fail' as '0'. Which of the following is a correct interpretation of odds of .05?&#10;A) The odds of passing are about equal to those of failing&#10;B) The odds of passing are greater than those of failing&#10;C) The odds of passing are less than those of failing&#10;D) Cannot be determined from the information provided

Accepted Answer

The odds of passing are very small, only about 5% odds (i.e., there is a small probability of passing).

Question 4

A log likelihood value of .50 suggests which one of the following?&#10;A) Bad model fit&#10;B) Neither good nor bad model fit&#10;C) Good model fit&#10;D) Cannot be determined from the information provided

Accepted Answer

The log likelihood value will approach zero the closer the likelihood value is to one. When this happens, this suggests the observed data could be generated from these population parameters. In other words, the smaller the log likelihood, the better the model fit.

Question 5

Which one of the following is NOT a tool for providing evidence of overall model fit?&#10;A) Box's M test&#10;B) change in log likelihood&#10;C) Hosmer and Lemeshow&#10;D) pseudo-variance explained

Accepted Answer

Several overall model tests can be used including: (a) change in log likelihood; (b) Hosmer and Lemeshow goodness of fit test; (c) pseudo-variance explained; and (d) predicted group membership.

Question 6

Which one of the following is NOT a tool for providing evidence of overall model fit?&#10;A) change in log likelihood&#10;B) Kolmogorov-Smirnov&#10;C) pseudo-variance explained&#10;D) predicted group membership

Accepted Answer

Several overall model tests can be used including: (a) change in log likelihood; (b) Hosmer and Lemeshow goodness-of-fit test; (c) pseudo-variance explained; and (d) predicted group membership.

Question 7

In a binary logistic model, 'receipt of services' is coded as '1' and 'nonreceipt' is coded as '0.' Of 40 cases coded as receiving services, only 10 cases were classified correctly. Of the 20 cases coded as not receiving services, 15 were classified correctly. What is the sensitivity in this situation?

A) 0%
B) 25%
C) 75%
D) 100%

Accepted Answer

Of those 40 cases who were coded as receiving services (i.e., coded as 1 for the dependent variable), 10 were correctly classified. Thus the sensitivity is 10/40 or about 25%.

Question 8

In a binary logistic model, 'receipt of services' is coded as '1' and 'nonreceipt' is coded as '0.' Of 40 cases coded as receiving services, only 10 cases were classified correctly. Of the 20 cases coded as not receiving services, 15 were classified correctly. What is the sensitivity in this situation?

A) 0%
B) 25%
C) 75%
D) 100%

Accepted Answer

Of those 20 cases who were coded as not receiving services (i.e., coded as 0 for the dependent variable), 15 were correctly classified. Thus the sensitivity is 15/20 or about 75%.

Question 9

A researcher finds a BIC value of 7.50. What can be interpreted by this?&#10;A) there is a large effect&#10;B) individual coefficient is statistically significantly different from zero and this is a strong difference&#10;C) the proportion of variance in the dependent variable accounted for by the model is about 75%

Accepted Answer

BIC values below zero suggest that the variable contributes little to the model. BIC values between 0 and +2 are considered weak; between 2 and 6, positive; between 6 and 10, strong; and more than 10, very strong.

Question 10

The method of predictor entry where the researcher allows the algorithm to select the sequence for the individual predictor variables is which one of the following?&#10;A) backward&#10;B) hierarchical regression&#10;C) simultaneous&#10;D) stepwise

Accepted Answer

Stepwise logistic regression is a data-driven model building technique where the computer algorithms drive variable entry rather than theory.

Deck 5: Logistic Regression