pow: Simulation of the type I error rate and the power

Description Usage Arguments Value Author(s) See Also

Description

This is an auxiliary function. It calculates the type I error rate and the power of the design with an internal pilot study for different timings of the pilot study and for multiple actual differences in expected differences. The originally planned sample size is calculated on the basis of an assumed standard deviation. A distinction is made between one-sided and two-sided tests.

Usage

1
2
pow(delta = 0, Delta, sd, test = 1, alpha = 0.05, beta = 0.2, prop = seq(0.1, 1, 0.05),
    adj = F, rule = F, nbound = 500, simu = 10000)

Arguments

delta

Number. Expectation difference of two samples.
If you select a Test for superiority/ difference then select 'delta = 0'.
Only if you select a Test for non-inferiority you can select 'delta != 0'.
Attention: If you chose 'test = 1' and 'delta != 0', the test for non-inferiority will
automatically be applied.
If not specified, delta is set to 0.

Delta

Number/ Vector of numbers. Relevant difference of expected values in the alternative hypothesis.

sd

Number. Assumed standard deviation of the data.
Used to calculate the originally planned number of cases.

test

Number. What type of hypothesis test should be performed. One-sided (Superiority/
Non-Inferiority test) or two-sided (Test for difference).
One-sided (test = 1): Superiortity H0: mu_x - mu_y <= 0 vs. H1: mu_x - mu_y > 0
Non-Inferiority H0: mu_x - mu_y >= delta vs. H1: mu_x - mu_y < delta
Two-sided (test = 2): Difference H0: |mu_x - mu_y| = 0 vs. H1: |mu_x - mu_y| != 0
Attention: Choice of delta. (see delta)
If not specified, the one-Sided Test (Superiority/ Non-Inferiority Test) is used.

alpha

Number. Desired alpha-level of the test.
If not specified, alpha is set to 0.05.

beta

Number. Acceptable beta error of the test.
If not specified, beta is set to 0.2.

prop

Number/ sequence of numbers.
Timing of the internal pilot study depending on the originally planned sample size.

adj

Logical. Should the one-sample variance, calculated in the internal pilot study, be adjusted?

rule

Logical. Should the sample size adjustment rule be applied by Wittes and Brittain?

nbound

Number. Upper limit of the sample size.
Attention: Only if you choose nbound can a suitable standard deviation range for the plots
be calculated automatically.
If no nbound are defined then a standard deviation range must be chosen (see sd_ber).

simu

Number. How many simulations should be performed?
If not specified, simu is set to 10000.

Value

This function only creates the power values for multiple timings of the internal pilot studies.
The output is used in the function pow_prop to visualize the power depending on the
timing for different true expected value differences.

Author(s)

Csilla van Lunteren

See Also

sim_calc


vanLunteren/plan_norm documentation built on May 20, 2019, 12:32 p.m.