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

The corresponding syntax, to run the logistic regression, create the new variable
PRE_1, and generate an ROC curve follows:

LOGISTIC REGRESSION VARIABLES fracture

/METHOD¼ENTER agecat head patellar flex weight

/SAVE¼PRED

/CLASSPLOT

/CRITERIA¼PIN(0.05) POUT(0.10) ITERATE(20) CUT(0.5).

ROC PRE_1 BY fracture (1)

/PLOT¼CURVE

/PRINT¼COORDINATES

/CRITERIA¼CUTOFF(INCLUDE) TESTPOS(LARGE) DISTRIBUTION(FREE) CI(95)

/MISSING¼EXCLUDE.

The output containing the parameter estimates of the logistic regression as well as
the resultant ROC curve from the model follow:

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

Step 1a agecat .556 .399 1.938 1 .164 1.744

head .218 .376 .337 1 .562 1.244

patellar .627 .352 3.175 1 .075 1.872

flex .528 .374 1.988 1 .159 1.695

weight 1.506 .409 13.532 1 .000 4.507

Constant 3.466 .412 70.837 1 .000 .031

aVariable(s) entered on step 1: agecat, head, patellar, flex, weight.

ROC Curve

1.0

0.8

0.6

Sensitivity0.4

1–specificity

Diagonal segments are produced by ties.

0.2

0.0

0.0 0.2 0.4 0.6 0.8 1.0

SPSS 641