Back calculating t-statistic in R

Question

I have these information from an independent samples t-test.

p-value=0.989

sample size for group 1 (n1)= 17

sample size for group 2 (n2)= 18

mean difference= 0.30

I need the t-statistic for this test. I did trial and error with the code for p-value to get the t-statistic, that is 0.014 2*pt(0.014, 33, lower.tail=F) #p=0.989

I need a the R code that gets the p-value and df as input, and provides 0.014 as the output.

You might have success looking at https://stats.stackexchange.com/ instead of stackoverflow, they specialize in these kinds of questions. — Max, May 05 '21 at 16:18

G. Grothendieck · Answer 1 · 2021-05-04T16:06:50.320

1

Find the root of the indicated function assuming that it lies between 0 and 6:

giving:

uniroot(function(x, p) p - 2 * pt(x, 33, lower.tail = FALSE), c(0, 6), p = 0.989)

giving:

$root
[1] 0.01390167

$f.root
[1] 7.863003e-06

$iter
[1] 3

$init.it
[1] NA

$estim.prec
[1] 6.103516e-05

Here is another example:

2 * pt(1.901367, 19, lower.tail = FALSE)
## [1] 0.07252959

uniroot(function(x, p) p - 2 * pt(x, 19, lower.tail = FALSE), c(0, 6), p = 0.07252959)

giving:

$root [1] 1.901365

$f.root [1] -2.672236e-07

$iter [1] 9

$init.it [1] NA

$estim.prec [1] 6.103516e-05

edited May 04 '21 at 16:06

answered May 04 '21 at 15:45

G. Grothendieck

254,981
17
203
341

I had changed that before your comment, not in response to it. – G. Grothendieck May 04 '21 at 16:48

score 0 · Accepted Answer · answered May 04 '21 at 18:26

0

There's no need of uniroot. Use qt:

> qt(0.989/2, df = 33, lower.tail = FALSE)
[1] 0.01389174

answered May 04 '21 at 18:26

Stéphane Laurent

75,186
15
119
225

Yes! I was looking for this. Thank you. – Jazmin2252 May 05 '21 at 21:39

Back calculating t-statistic in R

2 Answers2