htest support

[DominiqueMakowski]

This is a follow up to discuss the htest case.

[strengejacke]

Copied from PR:

See here: we have no information on parameter names:

str(t.test(extra ~ group, data = sleep))
#> List of 9
#>  $ statistic  : Named num -1.86
#>   ..- attr(*, "names")= chr "t"
#>  $ parameter  : Named num 17.8
#>   ..- attr(*, "names")= chr "df"
#>  $ p.value    : num 0.0794
#>  $ conf.int   : num [1:2] -3.365 0.205
#>   ..- attr(*, "conf.level")= num 0.95
#>  $ estimate   : Named num [1:2] 0.75 2.33
#>   ..- attr(*, "names")= chr [1:2] "mean in group 1" "mean in group 2"
#>  $ null.value : Named num 0
#>   ..- attr(*, "names")= chr "difference in means"
#>  $ alternative: chr "two.sided"
#>  $ method     : chr "Welch Two Sample t-test"
#>  $ data.name  : chr "extra by group"
#>  - attr(*, "class")= chr "htest"

Here, we only have a reference to variable names in a single character string, difficult to extract properly:

library(sjmisc)
data(efc)
str(t.test(efc$c12hour, efc$neg_c_7, data = sleep))
#> List of 9
#>  $ statistic  : Named num 18
#>   ..- attr(*, "names")= chr "t"
#>  $ parameter  : Named num 912
#>   ..- attr(*, "names")= chr "df"
#>  $ p.value    : num 2.98e-62
#>  $ conf.int   : num [1:2] 27.2 33.9
#>   ..- attr(*, "conf.level")= num 0.95
#>  $ estimate   : Named num [1:2] 42.4 11.8
#>   ..- attr(*, "names")= chr [1:2] "mean of x" "mean of y"
#>  $ null.value : Named num 0
#>   ..- attr(*, "names")= chr "difference in means"
#>  $ alternative: chr "two.sided"
#>  $ method     : chr "Welch Two Sample t-test"
#>  $ data.name  : chr "efc$c12hour and efc$neg_c_7"
#>  - attr(*, "class")= chr "htest"

[DominiqueMakowski]

maybe we can move it to 0.3.0 as it is not urgent?

[strengejacke]

Yes, I have little time the next days to work on packages, so I would focus on polishing docs and testing, in order to get a stable and nice release 0.2.0...

The htest-class is really horrible in terms of tidying.

[IndrajeetPatil]

Looks like we have added an extensive support for most of the htest and pairwise.htest objects.

I feel like only stats::oneway.test is currently not supported, which is the one I use in my packages 😅

Can we please support it? I won't need to use broom::tidy for any of the htest objects then.

[strengejacke]

Do we need oneway.test here, or in parameters? There isn't probably much to extract for insight.