The difference between Fischer's least significant difference (LSD) and Tukey's honestly significant difference (HSD) is that the former is more sensitive to small differences as it is based on a mean that is highly influenced by outliers than the latter that is based on the median. This makes the HSD more robust and reliable than LSD.
I heard about Duncan's multiple range test (DMRT), but I have no idea of its limitations and merits. Can someone help me understand the principles, merits and drawbacks of DMRT in comparison with HSD? Thank you.
Hi,
Here are few references:
Presentation 12 Duncan's multiple range test (DMRT)
http://www.stat.ucla.edu/~dinov/courses_students.dir/03/Spr/Stat110B.dir/STAT110B_notes.dir/lecture03_ANOVA.pdf
DMRT is more useful than the LSD when larger pairs of means are being compared, especially when those values are in a table. DMRT tends to require larger differences between means compared to the LSD, which guards against Type I error.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling