Multiple comparisons T-test function-CodePudding

I have the table with 9 collumn, 8 of which presented by numeric data and last collumn is grouping variable with 3 categories. It looks like in example of the table below.

data.frame(var1=rnorm(9), var2=rnorm(9), 
           var3=rnorm(9), var4=rnorm(9), 
           group=rep(c('A','B','C'),each=3))

I tried to create a function to calculate statistic and p.value of T-test in three pairs in each collumn with numeric data. I had no problem with creating a function that generates result in list format. The goal is to get a table format of results ( statistic (t) and p.value) for every pair in each collumn.

CodePudding user response：

set.seed(1)
df <- data.frame(
  var1 = rnorm(9),
  var2 = rnorm(9),
  var3 = rnorm(9),
  var4 = rnorm(9),
  group = rep(c('A', 'B', 'C'), each = 3)
)

df
#>         var1        var2        var3        var4 group
#> 1 -0.6264538 -0.30538839  0.82122120 -1.47075238     A
#> 2  0.1836433  1.51178117  0.59390132 -0.47815006     A
#> 3 -0.8356286  0.38984324  0.91897737  0.41794156     A
#> 4  1.5952808 -0.62124058  0.78213630  1.35867955     B
#> 5  0.3295078 -2.21469989  0.07456498 -0.10278773     B
#> 6 -0.8204684  1.12493092 -1.98935170  0.38767161     B
#> 7  0.4874291 -0.04493361  0.61982575 -0.05380504     C
#> 8  0.7383247 -0.01619026 -0.05612874 -1.37705956     C
#> 9  0.5757814  0.94383621 -0.15579551 -0.41499456     C

nm_grp <- unique(df$group)

library(tidyverse)

grp_split <- combn(nm_grp, m = 2) %>%
  data.frame() %>% 
  set_names(nm = map_chr(., paste0, collapse = ""))

l <- map(grp_split, ~ filter(df, group %in% .x))

res_list <- map(
  .x = l,
  .f = function(x)
    map(
      .x = select(x, where(is.numeric)),
      .f = function(y)
        t.test(y ~ x$group, data = x)
    )
)

res_nm <- map(names(grp_split), ~paste0(.x, ": ", names(df[-5]))) %>% flatten_chr()

map_df(flatten(res_list), broom::glance) %>% 
  add_column(nm = res_nm, .before = 1)
#> # A tibble: 12 x 11
#>    nm     estim~1 estim~2 estim~3 stati~4 p.value param~5 conf.~6 conf.~7 method
#>    <chr>    <dbl>   <dbl>   <dbl>   <dbl>   <dbl>   <dbl>   <dbl>   <dbl> <chr> 
#>  1 AB: v~  -0.794  -0.426   0.368  -1.04   0.381     2.76  -3.35    1.76  Welch~
#>  2 AB: v~   1.10    0.532  -0.570   1.00   0.388     3.10  -2.33    4.54  Welch~
#>  3 AB: v~   1.16    0.778  -0.378   1.38   0.298     2.05  -2.36    4.67  Welch~
#>  4 AB: v~  -1.06   -0.510   0.548  -1.52   0.206     3.79  -3.03    0.912 Welch~
#>  5 AC: v~  -1.03   -0.426   0.601  -3.21   0.0736    2.22  -2.28    0.224 Welch~
#>  6 AC: v~   0.238   0.532   0.294   0.383  0.725     3.32  -1.64    2.11  Welch~
#>  7 AC: v~   0.642   0.778   0.136   2.45   0.104     2.61  -0.267   1.55  Welch~
#>  8 AC: v~   0.105  -0.510  -0.615   0.156  0.884     3.64  -1.84    2.05  Welch~
#>  9 BC: v~  -0.232   0.368   0.601  -0.331  0.771     2.04  -3.19    2.72  Welch~
#> 10 BC: v~  -0.865  -0.570   0.294  -0.850  0.471     2.45  -4.56    2.83  Welch~
#> 11 BC: v~  -0.514  -0.378   0.136  -0.593  0.606     2.34  -3.77    2.74  Welch~
#> 12 BC: v~   1.16    0.548  -0.615   1.99   0.117     3.97  -0.461   2.79  Welch~
#> # ... with 1 more variable: alternative <chr>, and abbreviated variable names
#> #   1: estimate, 2: estimate1, 3: estimate2, 4: statistic, 5: parameter,
#> #   6: conf.low, 7: conf.high
```

<sup>Created on 2022-08-23 with [reprex v2.0.2](https://reprex.tidyverse.org)</sup>