test_auc_1curve: Power Analsis for Full Area Under the ROC Curve
In nutterb/junkyard: Package development playground

Description Usage Arguments Details Author(s) References Examples

Calculates the power, sample size, or other parameters for studies using Area Under the Curve (AUC) as the endpoint of an anlysis using a single continuous or ordinal predictor.

1
2
3

test_auc_1curve(auc0 = NULL, auc1 = NULL, n = NULL, power = NULL,
  alpha = 0.05, weights = list(c(1, 1)), one_tail = FALSE,
  ordinal = FALSE)

`auc0`	The hypothesize (null) value of AUC.
`auc1`	The observed or desired AUC.
`n`	The total sample size
`power`	The power of the test
`alpha`	The significance level of the test.
`weights`	Weighting of subjects with the condition and those without. A vector `1, 3` indicates there is one subject with the condition for every three subjects without it. Defaults to equal sizes. The weights are normalized, so they do not have to sum to one.
`one_tail`	Logical. Indicates if the test should be a one tailed test.
`ordinal`	Logical. Indicates if the independent variable in the study is ordinal. This is currently ignored.

The Hanley and McNeil estimator is used for continuous predictors. Obuchowski's estimator for discrete ratings is used for ordinal predictors.

Benjamin Nutter

Nancy A Obuchowski, "Sample size calculations in studies of test accuracy," Statistical Methods in Medical Research 1998; 7: 371-392

James A. Hanley and Barbara J. McNeil, "The Meaning and Use of the Area under a Receiver Operating Characteristic (ROC) Curve," Radiology. Vol 143. No 1, Pages 29-36, April 1982.

The bulk of this code is compared against C Zepp's SAS Macro ROCPOWER. See reference 25 in the Obuchowski paper above. http://www.bio.ri.ccf.org/doc/rocpower_help.txt http://www.bio.ri.ccf.org/doc/rocpower.sas

#* First example from \url{http://www.bio.ri.ccf.org/doc/rocpower_help.txt}
#* but solving for n
test_auc_1curve(auc0 = .75, auc1 = .92,
                n = NULL, alpha=.02,
                power=.893, weights = list(c(35, 50)))
#* First example from \url{http://www.bio.ri.ccf.org/doc/rocpower_help.txt}
test_auc_1curve(auc0 = .75, auc1 = .92,
                n = 85, alpha=.02,
                power=NULL,
                weights = list(c(35, 50)))