NOSTASOT: Finding the NOSTASOT dose

Description

The NOSTASOT dose is the No-Statistical-Significance-Of-Trend dose -- the largest dose at which no trend in the rate of response has been observed. It is often used to determine a safe dosage level for a potentially toxic compound.

Usage

NOSTASOT(
  cbdata,
  test = c("RS", "GEE", "GEEtrend", "GEEall", "SO"),
  exact = test == "SO",
  R = 100,
  sig.level = 0.05,
  control = soControl()
)

Value

a list with two components

NOSTASOT: character string identifying the NOSTASOT dose.
p: numeric vector of the p-values of the tests actually performed.

The last element corresponds to all doses included, and will not be missing. p-values for tests that were not actually performed due to the procedure stopping are set to NA.

Arguments

cbdata: a CBData object
test: character string defining the desired test statistic. See trend.test for details.
exact: logical, should an exact permutation test be performed. See trend.test for details.
R: integer, number of permutations for the exact test
sig.level: numeric between 0 and 1, significance level of the test
control: an optional list of control settings for the stochastic order ("SO") test, usually a call to soControl. See there for the names of the settable control values and their effect.

Author

Aniko Szabo, aszabo@mcw.edu

Details

A series of hypotheses about the presence of an increasing trend overall, with all but the last group, all but the last two groups, etc. are tested. Since this set of hypotheses forms a closed family, one can test these hypotheses in a step-down manner with the same sig.level type I error rate at each step and still control the family-wise error rate.

The NOSTASOT dose is the largest dose at which the trend is not statistically significant. If the trend test is not significant with all the groups included, the largest dose is the NOSTASOT dose. If the testing sequence goes down all the way to two groups, and a significant trend is still detected, the lowest dose is the NOSTASOT dose. This assumes that the lowest dose is a control group, and this convention might not be meaningful otherwise.

References

Tukey, J. W.; Ciminera, J. L. & Heyse, J. F. (1985) Testing the statistical certainty of a response to increasing doses of a drug. Biometrics 41, 295-301.

Examples

Run this code


data(shelltox)
NOSTASOT(shelltox, test="RS")

Run the code above in your browser using DataLab