Better Functions

library(scienceverse)
#> 
#> ************
#> Welcome to scienceverse For support and examples visit:
#> http://scienceverse.github.io/
#> - Get and set global package options with: scienceverse_options()
#> ************

Scienceverse provides some functions to replace common tests like t.test to provide data that is useful for meta-analyses. There is only one very basic example right now, but we will expand this in the future.

t.test vs t_test

The output of t.test does not return the number of data points per group. You can figure this out from the df for one-sample and paired t-tests, but not for two-sample t-tests. The function t_test returns a list with the same items as t.test plus n, which is the number of data points or pairs included in the one-sample or paired t-test (after excluding NAs, which t.test does silently), or a 2-number named vector of the number of data points in each group for a two-sample t-test.

Data

First, we’ll set up an example data frame with 20 rows, two continuous columns (x and y) and a grouping column (g).

dat <- data.frame(
  x = rnorm(20),
  y = rnorm(20, 0.5),
  g = rep(c("G1", "G2"), 10)
)

One sample

res <- t_test(dat$x)
res$n
#> [1] 20

Paired samples

res <- t_test(dat$x, dat$y, paired = TRUE)
res$n
#> [1] 20

Independent samples

Grp1 <- dat[dat$g== "G1", ]$x
Grp2 <- dat[dat$g== "G2", ]$x
res <- t_test(Grp1, Grp2)
res$n
#> Grp1 Grp2 
#>   10   10

The n is a named vector for independent-samples t-tests. The name comes from the variable names of x and y (not their original names in the data frame), unless you specify it directly with the names argument.

res <- t_test(Grp1, Grp2, names = c("A", "B"))
res$n
#>  A  B 
#> 10 10

You can also use the formula version.

res <- t_test(x ~ g, data = dat)
res$n
#> G1 G2 
#> 10 10