Timeline for box-cox transformation altered my anova result
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 23, 2018 at 17:31 | comment | added | kjetil b halvorsen♦ | In reality, boxcox transform ( if estimated in the usual likelihood way) will tend to stabilize variance rather than normalize. Stabilizing variance makes a much larger contribution to the likelihood. | |
| Feb 25, 2017 at 5:08 | vote | accept | upendra | ||
| Mar 28, 2015 at 0:21 | comment | added | upendra | townslay - thanks again for your help. I did the multiple anova and i did not find any difference between the lm + anova and multiple anova. Am i interpreting wrong here? Anyway i have edited my post again and now it make better read. Any further comments on this? Can i assume log-transformation is ok for my data? | |
| Mar 27, 2015 at 23:38 | comment | added | jeramy townsley | Ah--I reread your post now that you have edited it--it does look like you are using two dependent variables--conductance and photosynthesis. If you are testing them separately, with each of your independent variables, it's still just multiple ANOVA, and the "aov" commend, but if you want to test them together, then it becomes a MANOVA test | |
| Mar 27, 2015 at 23:35 | comment | added | jeramy townsley | Sorry--I made a comment earlier, but deleted it. I had referred originally to "multiple" anova, which is different from manova--the latter is called "multivariate" anova, which refers to having 2+ "dependent" variables--if "COND" or the respiration is your only dependent variable, then you would not use MANOVA. "Multiple" ANOVA uses 2+ "independent" variables, which is what you are showing us above. They are different commands in R: MANOVA actually uses "manova" while multiple anova simply uses "aov" statmethods.net/stats/anova.html | |
| Mar 27, 2015 at 23:10 | comment | added | upendra | you mean Manova? I have tried that too and i got the same result as earlier with ANOVA | |
| Mar 27, 2015 at 19:47 | comment | added | upendra | thanks a lot for your helpful comments and suggestions. Basically i am looking to see how the temperature, family, rank and genotype of the tree effects the respiration trait. I am not sure how i can check the distribution of dependant varaiable - Temp.group (35 & 45), family (6 families), rank (short and tall) and genotype (2 genotypes per family) since they are categorical. Can you throw some light on it... | |
| Mar 27, 2015 at 2:01 | history | answered | jeramy townsley | CC BY-SA 3.0 |