Dear All,
I have several charts with ~ 15 columns in each.
In the analysis of significance, how do we group these columns in letter-based style (a, ab, b, c, cd, d ... etc)?.. Is there any easy software to use?
Many thanks!
Maybe you can apply some multivariate techniques, such as K-means Cluster Analysis, which help you to group your data according their similarity.
Those letters are often called a compact letter display.
You start with a test that compares the treatments and reports p-values. Usually the test will compare each treatment to each other treatment. As examples, Tukey HSD or Dunn test could be used.
From there, you assign the letters so that any two treatments sharing a letter are not significantly different. This can be done by hand, although in some cases this can get tricky.
As a software solution, the cldList function in the rcompanion package † in R can be used.
In the following example, there are three treatments and three total comparisons. A test such as Tukey or Dunn had been run earlier to determine the p-values.
if(!require(rcompanion)){install.packages("rcompanion")}
Example = read.table(header=T, text="
Comparison p-value
'I - II ' 0.60
'I - III' 0.02
'II - III' 0.60
")
library(rcompanion)
cldList(p-value ~ Comparison, data = Example)
# Group Letter MonoLetter
# 1 I a a
# 2 II ab ab
# 3 III b b
_________________________
† Disclosure: I am the author of this package.
Dear Mazen
You can use SPSS software to analysis your data, it is easy software and needs little training.
Good Luck
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis