Code
library("serocalculator")
# See inst/examples/exm-compare_seroincidence.R for complete examplesCompare seroincidence rates between two groups
Description
Perform a two-sample z-test to compare seroincidence rates between two groups. Since we use maximum likelihood estimation (MLE) for each seroincidence estimate and estimates from different strata or data sets are uncorrelated, we can use a simple two-sample z-test using the Gaussian distribution. The standard error for the difference is computed by adding the estimated variances and taking the square root.
Usage
compare_seroincidence(x, y = NULL, coverage = 0.95, verbose = FALSE, ...)
## S3 method for class 'seroincidence'
compare_seroincidence(x, y = NULL, coverage = 0.95, verbose = FALSE, ...)
## S3 method for class 'seroincidence.by'
compare_seroincidence(x, y = NULL, coverage = 0.95, verbose = FALSE, ...)
Arguments
x
|
A “seroincidence” object from est_seroincidence() or a “seroincidence.by” object from est_seroincidence_by()
|
y
|
A “seroincidence” object from est_seroincidence() (optional if x is a “seroincidence.by” object)
|
coverage
|
Desired confidence interval coverage probability (default = 0.95) |
verbose
|
Logical indicating whether to print verbose messages (default = FALSE) |
…
|
Additional arguments (currently unused) |
Details
When comparing two single “seroincidence” objects, this function performs a two-sample z-test and returns results in the standard htest format.
When applied to a “seroincidence.by” object (stratified estimates), the function compares all pairs of strata and returns a nicely formatted table with point estimates for the difference in seroincidence, p-values, and confidence intervals.
The test statistic is computed as:
\(z = \frac{\lambda_1 - \lambda_2}{\sqrt{SE_1^2 + SE_2^2}}\)
where \(\lambda_1\) and \(\lambda_2\) are the estimated incidence rates, and \(SE_1\) and \(SE_2\) are their standard errors.
Value
-
When comparing two
“seroincidence”objects: An object of class“htest”containing the test statistic, p-value, confidence interval, and estimates. -
When applied to a
“seroincidence.by”object: Atibble::tibble()with columns for each pair of strata, the difference in incidence rates, standard error, z-statistic, p-value, and confidence interval bounds.
Methods (by class)
-
compare_seroincidence(seroincidence): Compare two single seroincidence estimates -
compare_seroincidence(seroincidence.by): Compare all pairs of stratified seroincidence estimates