chisq.pexplore: Explore power as a function of sample sie using a one- or...
In julianje/mcpa: Intuitive power analyses through monte carlo simulations

Description Usage Arguments Value See Also Examples

ttest.pexplore computes (via simulation) the power of an experiment that will be analyzed using a t-test for a range of sample sizes. Rather than taking a theoretical distribution, this function takes empirical data and bootstraps them to calculate the power. For an equivalent function that does not rely on pilot data see ttestpower.

1
2
3

chisq.pexplore(x, y = NULL, lown, topn, r = 10000,
  alternative = c("two.sided", "less", "greater"), mu = NULL,
  alpha = 0.05, plotit = TRUE)

`lown`	smallest sample size to explore.
`topn`	largest sample size to explore.
`r`	number of simulations to compute power.
`alternative`	type of alternative hypothesis in binomial test. Must be "`two.sided`" (default), "`greater`", or "`less`".
`mu`	mean value according to null hypothesis (default = `0`). Only used in one sample t-tests.
`alpha`	significance threshhold.
`plotit`	logical (default=`TRUE`) value. Function generates a plot when `TRUE` and returns a data frame otherwise.

The probability of finding p < α with the experiment description.

ttest.pow, ttest.ppow, ttest.explore, and ttest.pexplore.

1
2
3

ttest.pexplore(x=c(0, 5, 10), lown=16, topn=24) # Power for a one-sample t-test with n in 16-24. Pilot data consists of three data points.
ttest.pexplore(x=c(0, 5, 10), lown=16, topn=24,mu = -5) # Same as above, changing the avarege under the null to -5.
ttest.pexplore(x=c(0, 5, 10), lown=16, topn=24, y=c(9, 3, 2, 1)) # Power for a two-sample t-test with n=16-24 (per condition) using unbalanced pilot data.