1–2*alpha confidence interval given point estimate, CV, and n

Utility function to calculate the 1–2α CI given point estimate, CV, and n for the various designs covered in this package.

CI.BE(alpha = 0.05, pe, CV, n, design = "2x2", robust = FALSE)

Arguments

alpha

Type I error probability, significance level. Defaults to 0.05.

pe

Point estimate (GMR).

CV

Coefficient of variation as ratio (not percent).

n

Total number of subjects if a scalar is given.
Number of subjects in (sequence) groups if given as vector.

design

Character string describing the study’s design.
See known.designs() for designs covered in this package.

robust

Defaults to FALSE.
Setting to TRUE will use the degrees of freedom according to the ‘robust’ evaluation (aka Senn’s basic estimator). These degrees of freedom are calculated as n-seq.
See known.designs()$df2 for designs covered in this package.

Value

Returns the 1–2α confidence interval.
Returns a vector with named elements lower, upper if arguments pe and CV are scalars, else a matrix with columns lower, upper is returned.

Note

The function assumes an evaluation using log-transformed data.
The function assumes equal variances in case of design="parallel" and the higher order crossover designs.
The implemented formula covers balanced and unbalanced designs.

Whether the function vectorizes properly is not thoroughly tested.

Author

D. Labes

Examples

# 90% confidence interval for the 2x2 crossover
# n(total) = 24
CI.BE(pe = 0.95, CV = 0.3, n = 24)
#>     lower     upper 
#> 0.8213465 1.0988055 
# should give
#     lower     upper 
# 0.8213465 1.0988055 
# same total number but unequal sequences
CI.BE(pe = 0.95, CV = 0.3, n = c(13, 11))
#>     lower     upper 
#> 0.8209294 1.0993637 
#     lower     upper
# 0.8209294 1.0993637