Logistic Regression: A Self-learning Text, Third Edition (Statistics in the Health Sciences)

1. For a discrimination cut-point of0.300in the Classifi-
   cation Table provided above,
   a. fill in the table below to show the cell frequencies
   for the number of true positives (nTP), false posi-
   tives (nFP), true negatives (nTN), and false negatives
   (nFN):

True (Observed) Outcome

cp¼0.30 Y¼ 1 Y¼ 0

Predicted Y¼ 1 nTP¼ nFP¼

Outcome Y¼ 0 nFN¼ nTN¼

n 1 ¼ 114 n 0 ¼ 175

b. Using the cell frequencies in the table of part 1a,

computein percentagesthe sensitivity, specificity,

1 specificity, false positive, and false negative

values, and verify that these results are identical

to the results shown in the Classification Table for

cut-point 0.300:

Sensitivity %¼

Specificity %¼

1 specificity %¼

False positive %¼

False negative %¼

c. Why are the 1specificity and false positive per-

centagesnotidentical even though they both use

the (same) number of false positive subjects in

their calculation?

d. How is the value of 70.9 in the column labeled

“Correct” computed and how can this value be

interpreted?

e. How do you interpret values for sensitivity and

specificity obtained for the cut-point of 0.300 in

terms of how well the model discriminates cases

from noncases?

f. What is the drawback to (exclusively) using the

results for the cut-point of 0.300 to determine how

well the model discriminates cases from noncases?

2. Using the following graph, plot the points on the graph
   that would give the portion of the ROC curve that
   corresponds to the following cut-points: 0.000, 0.200,
   0.400, 0.600, 0.800, and 1.000

382 10. Assessing Discriminatory Performance of a Binary Logistic Model