Skip to content

Games-Howell function for post-hoc multiple comparisons

1 message · Steven Yen

Steven Yen
# 
Dear R users,

I am conducting multiple comparisons among 12 groups (after a significant
F-test) that are heteroscedastic (as judged by a significant Levene's test).
It seems from the literature that the Games-Howell post-hoc test is the most
appropriate for these data - but, I can't seem to locate a function on the
R-site or CRAN or the help lists that implements this test in any R package.
I'm unsure whether the TukeyHSD test is tolerant of heteroscedasticity - I
think not, and I know Hochberg's isn't. Does anyone know of a test available
in R that is suitable for these data? Thanks a lot.

Steve