Auxiliary functions to (back) transform effect estimates or confidence / prediction interval limit(s).
transf(x, sm, func = NULL, args = NULL)cor2z(x)
p2asin(x)
p2logit(x)
VE2logVR(x)
backtransf(x, sm, n, time, func = NULL, args = NULL)
asin2ir(x, time = NULL)
asin2p(x, n = NULL)
logit2p(x)
logVR2VE(x)
z2cor(x)
Numerical vector with effect estimates, lower or upper confidence / prediction interval limit(s).
Summary measure.
User-specified function for (back) transformation.
Function arguments for user-specified function.
Sample size(s) to back transform Freeman-Tukey transformed proportions.
Time(s) to back transform Freeman-Tukey transformed incidence rates.
Guido Schwarzer guido.schwarzer@uniklinik-freiburg.de
Often in a meta-analysis, effect estimates are transformed before
calculating a weighted average. For example, the log odds ratio and
its standard error is used instead of the odds ratio in R function
metagen. To report the results of a meta-analysis,
effect estimates are typically back transformed to the original
scale. R package meta provides some auxiliary functions for
(back) transformations.
The following auxiliary functions are provided by R package meta to transform effect estimates or confidence / prediction interval limits.
| Function | Transformation |
cor2z | Correlations to Fisher's Z transformed correlations |
p2logit | Proportions to logit transformed proportions |
p2asin | Proportions to arcsine transformed proportions |
VE2logVR | Vaccine efficacy / effectiveness to log vaccine ratio |
Note, no function for the Freeman-Tukey arcsine transformation is provided as this transformation is based on the number of events and sample sizes instead of the effect estimates.
R function transf is a wrapper function for the above and
additional transformations, e.g., the log transformation using
log for odds or risk ratios. Argument sm
is mandatory to specify the requested transformation. It is also
possible to specify a different function with arguments func
and args.
The following auxiliary functions are available to back transform effect estimates or confidence / prediction interval limits.
| Function | Transformation |
asin2ir | Freeman-Tukey arcsine transformed rates to rates |
asin2p | (Freeman-Tukey) arcsine transformed proportions to proportions |
logit2p | Logit transformed proportions to proportions |
logVR2VE | Log vaccine ratio to vaccine efficacy / effectiveness |
z2cor | Fisher's Z transformed correlations to correlations |
Argument time is mandatory in R function asin2ir.
If argument n is provided in R function asin2p,
Freeman-Tukey arcsine transformed proportions are
back transformed. Otherwise, arcsine transformed proportions are
back transformed.
R function backtransf is a wrapper function for the above
and additional transformations, e.g., the exponential
transformation using exp for log odds or log
risk ratios. Argument sm is mandatory to specify the
requested transformation. For the Freeman-Tukey transformations,
argument n or time is mandatory.
It is also possible to specify a different function with arguments
func and args.
meta-sm