Hello. I am considering how to statistically compare 3 datasets (plants from different proveniences) consisted of measurements of leaf damage 30 leaves each provenience. I measured the damaged areas, but because leaves are not the same size I counted foliage damage in % from total area of the leave. Now I need to statistically compare what provenience is more damaged. Thank you for advices.
Hello Zuzana Janosikova Heckova. There are regression models for proportional (or fractional) outcome variables. You have not said what software you have. Using Stata, I would likely use either -fracreg- or -betareg-. You can see more info about both of them in this conference presentation:
- https://www.stata.com/meeting/mexico16/slides/Mexico16_Dorta.pdf
- https://www.stata.com/manuals/rfracreg.pdf
- https://www.stata.com/manuals/rbetareg.pdf
If you say what software you use, others may be able to direct you to the command(s) to use. HTH.
Because you have a setup analogous to a one-way analysis of variance (anova), you may have many options depending on the nature of your data and the software you are using. A traditional approach like anova or Kruskal-Wallis may make sense, or you might want to use a permutation test. As Bruce Weaver notes, beta regression also appears to be an option. It helps to start by plotting the data for the three provinces, to see the distributions of the data. Also, you might consider seriously what hypothesis you want to test. Are you committed to looking at means? Or would looking at medians be applicable? Or if one area is likely to have higher values than another?
You can use one way anova and complete the test by using Duncan multiple range test or Tukey. Anova will explain to you if there is a significant difference between the groups. If there is a significant difference, Duncan will explain which group is different.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biochemical Techniques
- Analytical Chemistry
- pH