ROC Analysis: Options

You can specify the following options for your ROC analysis:

Classification

Allows you to specify whether the cutoff value should be included or excluded when making a positive classification. This setting currently has no effect on the output.

Test Direction

Provides options for specifying which direction of the test result variable indicates increasing strength of conviction that the subject is test positive.

Parameters for Standard Error of Area

Allows you to specify the method of estimating the standard error of the area under the curve. Available methods are nonparametric and bi-negative exponential. The default Nonparametric setting provides estimates under the nonparametric assumption. The Bi-negative exponential setting provides estimates under the bi-negative exponential distribution assumption.

The section also allows you to specify the confidence level for the two-sided asymptotic confidence interval of the AUC. The available range is 0.0% to 100.0% (the default value is 95%).

Note: The setting only applies to the independent-group design and has no effect in the paired-sample design.

Missing Values

Allows you to specify how missing values are handled. When the setting is not selected, both user-missing values and system-missing values are excluded. When the setting is selected, user-missing values as treated as valid, system-missing values are excluded. Cases with system-missing values, in either the test variable or the state variable, are always excluded from the analysis.