Suggestion: support for t-tests

[cschwem2er]

Hi,

as sd-tests and t-tests in R do not provide a lot of information, adding a table function in sjPlot would be very useful.

This is the output of a paired t-test in Stata (see #189): It includes everything that is of interest (group specific and difference wise), means, standard deviations, confidence intervals, nr. of observations, df's, p values. In comparison this is the t-test in R:

For independent two sample t-tests, there could either be a separate function to check for equal variances (var.test in base R), or the t-test function could have a parameter, e.g.var.equal, with options auto, TRUE, or FALSE. In the case of auto, the function could automatically check for variance equality and adjust the degrees of freedom computation if necessary.

[jemus42]

In the meantime, you can pipe that t.test output into broom::tidy, which should give you all the information you might need. You can then pretty up the output via sjt.df or pixiedust, as I do in tadaatoolbox::tadaa_t.test.

[strengejacke]

Do you have a reproducible example that resembles the Stata output?

[cschwem2er]

Hi, should be easier with the efc dataset:

library(sjmisc)
data(efc)
t.test(c82cop1 ~ e16sex, data = efc, var.equal = TRUE)

	Two Sample t-test

data:  c82cop1 by e16sex
t = 0.11281, df = 898, p-value = 0.9102
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -0.07671384  0.08607089
sample estimates:
mean in group 1 mean in group 2 
       3.122034        3.117355