The 'multcompLetters' function in the R package 'multcompView' automates
a "grouping letters" algorithm documented in the following publication:
Piepho, Hans-Peter (2004) "An Algorithm for a Letter-Based
Representation of All-Pairwise Comparisons", Journal of
Computational and Graphical Statistics, 13(2)456-466.
hope this helps.
Spencer Graves
Data Analytics Corp. wrote:
Good morning,
I'm trying to write a function to assign the mean separation grouping
letters to the factors of an experiment. A client wants this so I
thought I'd write my own function for future use. But I'm having
trouble doing it because I don't understand the logic of the
assignment. As a general example, consider the following data and the
results from SAS and its weak cousin, JMP:
The data:
=======
Strain Nitrogen
3DOK1 19.40
3DOK1 32.60
3DOK1 27.00
3DOK1 32.10
3DOK1 33.00
3DOK5 17.70
3DOK5 24.80
3DOK5 27.90
3DOK5 25.20
3DOK5 24.30
3DOK4 17.00
3DOK4 19.40
3DOK4 9.10
3DOK4 11.90
3DOK4 15.80
3DOK7 20.70
3DOK7 21.00
3DOK7 20.50
3DOK7 18.80
3DOK7 18.60
3DOK13 14.30
3DOK13 14.40
3DOK13 11.80
3DOK13 11.60
3DOK13 14.20
COMPOS 17.30
COMPOS 19.40
COMPOS 19.10
COMPOS 16.90
COMPOS 20.80
Results from SAS:
Tukey Grouping Mean N Strain
A 28.820 5 3DOK1
B A 23.980 5 3DOK5
B C 19.920 5 3DOK7
B C 18.700 5 COMPOS
C 14.640 5 3DOK4
C 13.260 5 3DOK13
From JMP:
3DOK1 A 28.82
3DOK5 A B 23.98
3DOK7 B C 19.92
COMPOS B C 18.7
3DOK4 C 14.64
3DOK13 C 13.26
The results are the same as they should be. What I don't understand
is why only letters A-C? Why not D and E also since each of five
Strains appears in the first position of the 15 pairs?
Does anyone have a function to do this?
Thanks,
Walt Paczkowski
--------------------------------------------------------------------
This message was distributed by s-news@lists.biostat.wustl.edu. To
unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
the BODY of the message: unsubscribe s-news
|