help
help > RE: Effect Sizes using TFCE
Oct 24, 2021 11:10 PM | Alfonso Nieto-Castanon - Boston University
RE: Effect Sizes using TFCE
Hi Mayron,
Right, almost exactly, but a couple of small corrections:
a) in the original formula I believe the relationship for an arbitrarily-sized two-sample t-test was really
Cohen's d = T/sqrt(dof) * (N1+N2-2)/sqrt(N1*N2)
and b) in this case, when having additional groups (not-tested in your contrast but included in the model), I believe the formula should be (very similar to what you were suggesting):
Cohen's d = T/sqrt(dof) * sqrt((N-M)*(1/N1+1/N2))
where N is the total sample size across all groups (e.g. N=223 in this case), M is the number of groups (e.g. M=4), and N1 N2 are the sizes of the two groups being compared (e.g. N1=115 and N2=38)
Hope this helps
Alfonso
Originally posted by Mayron Pereira Picolo Ribeiro:
Right, almost exactly, but a couple of small corrections:
a) in the original formula I believe the relationship for an arbitrarily-sized two-sample t-test was really
Cohen's d = T/sqrt(dof) * (N1+N2-2)/sqrt(N1*N2)
and b) in this case, when having additional groups (not-tested in your contrast but included in the model), I believe the formula should be (very similar to what you were suggesting):
Cohen's d = T/sqrt(dof) * sqrt((N-M)*(1/N1+1/N2))
where N is the total sample size across all groups (e.g. N=223 in this case), M is the number of groups (e.g. M=4), and N1 N2 are the sizes of the two groups being compared (e.g. N1=115 and N2=38)
Hope this helps
Alfonso
Originally posted by Mayron Pereira Picolo Ribeiro:
Just following up on this...
Originally posted by Mayron Pereira Picolo Ribeiro:
Originally posted by Mayron Pereira Picolo Ribeiro:
Thank you so much, Alfonso!
Another follow-up question I have is related to transforming the T-stats to Cohen's d. I saw you mentioned the following on a previous post:
"Cohen's d, in your case that can be easily computed from your analysis T-stats and dofs as d = T / sqrt(dof)"
https://www.nitrc.org/forum/message.php?...
Even though I had 4 groups (n=223) in my analysis, I'd like to report effect sizes for the comparison of 2 groups (n=115 | 38 in one group and 77 in another). Individuals from groups that are not being compared here had 0s in both sequences.
In my effect size table, I get the df 216 (T(216)), but since the comparison I am focusing on n=115, should I divide the T-stat for 108 instead?
115 - 7 (contrast above 1 -1 0 0 0 0 0)
Thank you so much
mayron
Another follow-up question I have is related to transforming the T-stats to Cohen's d. I saw you mentioned the following on a previous post:
"Cohen's d, in your case that can be easily computed from your analysis T-stats and dofs as d = T / sqrt(dof)"
https://www.nitrc.org/forum/message.php?...
Even though I had 4 groups (n=223) in my analysis, I'd like to report effect sizes for the comparison of 2 groups (n=115 | 38 in one group and 77 in another). Individuals from groups that are not being compared here had 0s in both sequences.
In my effect size table, I get the df 216 (T(216)), but since the comparison I am focusing on n=115, should I divide the T-stat for 108 instead?
115 - 7 (contrast above 1 -1 0 0 0 0 0)
Thank you so much
mayron
Threaded View
| Title | Author | Date |
|---|---|---|
| Mayron Pereira Picolo Ribeiro | Oct 13, 2021 | |
| Alfonso Nieto-Castanon | Oct 13, 2021 | |
| Mayron Pereira Picolo Ribeiro | Oct 14, 2021 | |
| Mayron Pereira Picolo Ribeiro | Oct 24, 2021 | |
| Alfonso Nieto-Castanon | Oct 24, 2021 | |
| Mayron Pereira Picolo Ribeiro | Jan 31, 2022 | |
| Alfonso Nieto-Castanon | Feb 7, 2022 | |