pairwise_t_test() doesn't return statistic and estimate

Question

I am a beginner in R and do not understand why pairwise_t_test() from the rstatix package does not show the t-value and cohen's d.

pairwise_t_test(data = filtered_normdata, BZO_RWgesamt ~ Alter, p.adjust.method = "none")
# A tibble: 36 × 9
   .y.       group1 group2    n1    n2        p p.signif    p.adj p.adj.signif
 * <chr>     <chr>  <chr>  <int> <int>    <dbl> <chr>       <dbl> <chr>       
 1 BZO_RWge… 11     12        95    90 4.26e- 2 *        4.26e- 2 *           
 2 BZO_RWge… 11     13        95    90 6.17e- 5 ****     6.17e- 5 ****        
 3 BZO_RWge… 12     13        90    90 4.85e- 2 *        4.85e- 2 *           
 4 BZO_RWge… 11     14        95    74 4.85e- 9 ****     4.85e- 9 ****        
 5 BZO_RWge… 12     14        90    74 8.21e- 5 ****     8.21e- 5 ****        
 6 BZO_RWge… 13     14        90    74 3.72e- 2 *        3.72e- 2 *           
 7 BZO_RWge… 11     15        95    68 8.39e-18 ****     8.39e-18 ****        
 8 BZO_RWge… 12     15        90    68 1.22e-11 ****     1.22e-11 ****        
 9 BZO_RWge… 13     15        90    68 5   e- 7 ****     5   e- 7 ****        
10 BZO_RWge… 14     15        74    68 3.73e- 3 **       3.73e- 3 **          
# ℹ 26 more rows
# ℹ Use `print(n = ...)` to see more rows

I already tried to get the statistic and estimate like this:

BZO_H2_w <- pairwise_t_test(data = filtered_normdata, BZO_RWgesamt ~ Alter, p.adjust.method = "none")

BZO_H2_w$estimate and BZO_H2_w$statistic

but it didn't work. I would be really thankful if someone can help me :)

Answer 1

If it's the function from rstatix that you are asking about then it only seems to return the full detail of the tests if you request not to use a pooled standard deviation for the t-tests (pool.sd=FALSE).

If you look at the code, pairwise_t_test calls the R function stats::pairwise.t.test when you use a pooled SD, but uses a repeated application of stats::t.test when pool.sd=FALSE. It then uses broom::tidy to collate the statistics into a table.

So I don't think there's a good statistical reason for the difference in reporting, it just relies on different base functions, and while the R function t.test does return the t-statistic etc, pairwise.t.test only returns a matrix of p-values.

You can get the detail of the pairwise t-tests with pooled SD if you use lm to get a linear model (ANOVA) then emmeans to get the pairwise contrasts. Note the p-values from emmeans are the same as from pairwise.t.test with the pooled SD, but it returns the estimate and statistic as well.

So, with some toy data:

> dat <- data.frame(x=rep(1:10,10), y=rnorm(100))
> 
> rstatix::pairwise_t_test(dat, y~x, detailed=TRUE)

# A tibble: 45 × 10
   .y.   group1 group2    n1    n2     p method p.adj p.signif p.adj.signif
 * <chr> <chr>  <chr>  <int> <int> <dbl> <chr>  <dbl> <chr>    <chr>       
 1 y     1      2         10    10 0.536 T-test     1 ns       ns          
 2 y     1      3         10    10 0.453 T-test     1 ns       ns          
 3 y     2      3         10    10 0.896 T-test     1 ns       ns          
 4 y     1      4         10    10 0.427 T-test     1 ns       ns          
 5 y     2      4         10    10 0.86  T-test     1 ns       ns          
 6 y     3      4         10    10 0.964 T-test     1 ns       ns          
 7 y     1      5         10    10 0.781 T-test     1 ns       ns          
 8 y     2      5         10    10 0.733 T-test     1 ns       ns          
 9 y     3      5         10    10 0.636 T-test     1 ns       ns          
10 y     4      5         10    10 0.605 T-test     1 ns       ns          
# ℹ 35 more rows
# ℹ Use `print(n = ...)` to see more rows

> rstatix::pairwise_t_test(dat, y~x, pool.sd=FALSE, detailed = TRUE)

# A tibble: 45 × 17
   estimate estimate1 estimate2 .y.   group1 group2    n1    n2 statistic     p    df conf.low conf.high method alternative p.adj p.adj.signif
 *    <dbl>     <dbl>     <dbl> <chr> <chr>  <chr>  <int> <int>     <dbl> <dbl> <dbl>    <dbl>     <dbl> <chr>  <chr>       <dbl> <chr>       
 1  -0.276    -0.318   -0.0417  y     1      2         10    10   -0.725  0.479  15.6   -1.08      0.533 T-test two.sided       1 ns          
 2  -0.334    -0.318    0.0166  y     1      3         10    10   -0.760  0.457  18.0   -1.26      0.590 T-test two.sided       1 ns          
 3  -0.354    -0.318    0.0367  y     1      4         10    10   -0.782  0.445  18.0   -1.31      0.598 T-test two.sided       1 ns          
 4  -0.124    -0.318   -0.194   y     1      5         10    10   -0.306  0.763  17.0   -0.977     0.729 T-test two.sided       1 ns          
 5  -0.209    -0.318   -0.109   y     1      6         10    10   -0.458  0.653  18.0   -1.17      0.749 T-test two.sided       1 ns          
 6  -0.0428   -0.318   -0.275   y     1      7         10    10   -0.104  0.918  17.3   -0.909     0.823 T-test two.sided       1 ns          
 7   0.0437   -0.318   -0.361   y     1      8         10    10    0.0933 0.927  17.9   -0.940     1.03  T-test two.sided       1 ns          
 8  -0.150    -0.318   -0.167   y     1      9         10    10   -0.320  0.752  17.9   -1.14      0.835 T-test two.sided       1 ns          
 9  -0.321    -0.318    0.00292 y     1      10        10    10   -0.616  0.546  16.9   -1.42      0.778 T-test two.sided       1 ns          
10  -0.0583   -0.0417   0.0166  y     2      3         10    10   -0.158  0.876  16.0   -0.840     0.724 T-test two.sided       1 ns          
# ℹ 35 more rows
# ℹ Use `print(n = ...)` to see more rows

> mod1 <- lm(data=dat , y~factor(x)) 
> emmeans::emmeans(mod1, pairwise ~ x, adjust="none", infer=c(TRUE, TRUE))$contrast

 contrast estimate    SE df lower.CL upper.CL t.ratio p.value
 x1 - x2   -0.2759 0.444 90   -1.158    0.606  -0.622  0.5357
 x1 - x3   -0.3343 0.444 90   -1.216    0.547  -0.753  0.4533
 x1 - x4   -0.3543 0.444 90   -1.236    0.527  -0.798  0.4268
 x1 - x5   -0.1238 0.444 90   -1.006    0.758  -0.279  0.7809
 x1 - x6   -0.2086 0.444 90   -1.090    0.673  -0.470  0.6395
 x1 - x7   -0.0428 0.444 90   -0.925    0.839  -0.096  0.9234
 x1 - x8    0.0437 0.444 90   -0.838    0.925   0.098  0.9218
 x1 - x9   -0.1503 0.444 90   -1.032    0.731  -0.339  0.7357
 x1 - x10  -0.3206 0.444 90   -1.202    0.561  -0.722  0.4720
 x2 - x3   -0.0583 0.444 90   -0.940    0.823  -0.131  0.8957
 x2 - x4   -0.0784 0.444 90   -0.960    0.803  -0.177  0.8602
 x2 - x5    0.1521 0.444 90   -0.730    1.034   0.343  0.7326
 x2 - x6    0.0673 0.444 90   -0.814    0.949   0.152  0.8798
 x2 - x7    0.2331 0.444 90   -0.649    1.115   0.525  0.6007
 x2 - x8    0.3196 0.444 90   -0.562    1.201   0.720  0.4733
 x2 - x9    0.1257 0.444 90   -0.756    1.007   0.283  0.7777
 x2 - x10  -0.0446 0.444 90   -0.926    0.837  -0.101  0.9201
 x3 - x4   -0.0201 0.444 90   -0.902    0.862  -0.045  0.9640
 x3 - x5    0.2105 0.444 90   -0.671    1.092   0.474  0.6365
 x3 - x6    0.1257 0.444 90   -0.756    1.007   0.283  0.7777
 x3 - x7    0.2915 0.444 90   -0.590    1.173   0.657  0.5130
 x3 - x8    0.3779 0.444 90   -0.504    1.260   0.852  0.3967
 x3 - x9    0.1840 0.444 90   -0.698    1.066   0.415  0.6794
 x3 - x10   0.0137 0.444 90   -0.868    0.895   0.031  0.9754
 x4 - x5    0.2305 0.444 90   -0.651    1.112   0.519  0.6047
 x4 - x6    0.1457 0.444 90   -0.736    1.027   0.328  0.7434
 x4 - x7    0.3116 0.444 90   -0.570    1.193   0.702  0.4845
 x4 - x8    0.3980 0.444 90   -0.484    1.280   0.897  0.3722
 x4 - x9    0.2041 0.444 90   -0.678    1.086   0.460  0.6468
 x4 - x10   0.0338 0.444 90   -0.848    0.915   0.076  0.9395
 x5 - x6   -0.0848 0.444 90   -0.967    0.797  -0.191  0.8489
 x5 - x7    0.0810 0.444 90   -0.801    0.963   0.183  0.8556
 x5 - x8    0.1675 0.444 90   -0.714    1.049   0.377  0.7068
 x5 - x9   -0.0265 0.444 90   -0.908    0.855  -0.060  0.9526
 x5 - x10  -0.1968 0.444 90   -1.078    0.685  -0.443  0.6586
 x6 - x7    0.1658 0.444 90   -0.716    1.048   0.374  0.7096
 x6 - x8    0.2523 0.444 90   -0.629    1.134   0.568  0.5712
 x6 - x9    0.0583 0.444 90   -0.823    0.940   0.131  0.8957
 x6 - x10  -0.1120 0.444 90   -0.994    0.770  -0.252  0.8014
 x7 - x8    0.0865 0.444 90   -0.795    0.968   0.195  0.8460
 x7 - x9   -0.1075 0.444 90   -0.989    0.774  -0.242  0.8092
 x7 - x10  -0.2778 0.444 90   -1.160    0.604  -0.626  0.5330
 x8 - x9   -0.1939 0.444 90   -1.076    0.688  -0.437  0.6632
 x8 - x10  -0.3642 0.444 90   -1.246    0.517  -0.821  0.4140
 x9 - x10  -0.1703 0.444 90   -1.052    0.711  -0.384  0.7021