rss.AUC.test {generalRSS}R Documentation

RSS empirical likelihood ratio (ELR) test in two-sample comparison

Description

The rss.AUC.test function conducts an empirical likelihood ratio test to compare the Area Under the Curve (AUC) between two groups using RSS. It supports both balanced and unbalanced RSS designs.

Usage

rss.AUC.test(data1, data2, alpha = 0.05, delta0 = 0.5)

Arguments

data1

A numeric data frame of ranked set samples with columns rank for ranks and y for data values.

data2

A numeric data frame of ranked set samples with columns rank for ranks and y for data values.

alpha

A numeric value specifying the confidence level for the interval.

delta0

A numeric value indicating the hypothesized value of the AUC.

Details

This function performs an empirical likelihood ratio test to compare the Area Under the Curve (AUC) between two groups using Ranked Set Sampling (RSS). The test is equivalent to the Mann-Whitney U test, which evaluates whether there is a significant difference between the distributions of two groups. The function supports both Balanced RSS (BRSS) and unbalanced RSS (URSS), as described by Moon et al. (2022). Given two data frames of RSS data (data1 and data2) with rank and y columns, the function calculates the empirical likelihood ratio test statistic, confidence interval, and p-value based on the hypothesized AUC value delta0. Test for delta0=0.5 corresponds to the null hypothesis of the Mann-Whitney U test, which asserts that there is no difference between the two groups, implying that any randomly selected observation from one group is just as likely to be greater or smaller than an observation from the other group.

Value

RSS_AUC

The RSS AUC estimate.

CI

The confidence interval for the AUC.

-2*Log.LR

The empirical log likelihood ratio test statistics.

p.value

The p-value for the test.

References

C. Moon, X. Wang, and J. Lim. (2022) Empirical likelihood inference for area under the receiver operating characteristic curve using ranked set samples. Pharmaceutical Statistics, 21(6), 1219–1245.

See Also

rss.simulation: used for simulating Ranked Set Samples (RSS), which can serve as input.

rss.sampling: used for sampling Ranked Set Samples (RSS) from a population data set, providing input data.

Examples

## balanced RSS with a set size 3 and different sample sizes of 6 or 8 for each stratum,
## using imperfect ranking from a normal distribution with a mean of 0.
rss.data1<-rss.simulation(H=3,nsamp=c(6,6,6),dist="normal", rho=0.8,delta=0)
rss.data2<-rss.simulation(H=3,nsamp=c(8,8,8),dist="normal", rho=0.8,delta=0.2)

## RSS empirical likelihood ratio test
rss.AUC.test(data1=rss.data1, data2=rss.data2, alpha=0.05, delta0=0.5)

## Unbalanced RSS with a set size 3 and different sample sizes of 6, 10, and 8 for each stratum,
## using imperfect ranking from a normal distribution with a mean of 0.
rss.data1<-rss.simulation(H=3,nsamp=c(6,10,8),dist="normal", rho=0.8,delta=0)
rss.data2<-rss.simulation(H=3,nsamp=c(6,10,8),dist="normal", rho=0.8,delta=0.2)

## RSS empirical likelihood ratio test
rss.AUC.test(data1=rss.data1, data2=rss.data2, alpha=0.05, delta0=0.5)


[Package generalRSS version 0.1.3 Index]