Hello,
I'm trying to do statistics for my experiment - I have 4 fungi isolates and I cultured them in different concentrations of Cu (0, 50, 100, 250 uM). After 3 days I dried and weighted the biomass. I want to know which one of the isolates grows the best in which Cu concentration.
I'm trying to do statistics in R software. I did the normality test using the command
> shapiro.test(Cu1$BIOMASS) and I also did Bartlett test of homogenity of variances ( > bartlett.test(BIOMASS ~ interaction(CU_CONC,TRICH), data = Cu1)), and they came out right.
Now I wanted to do Two-way ANOVA, because the biomass would be affected by the fungi isolate and the concentration. But I'm not sure what commands should I use...
I kindly ask for your help.
Thank you
Here are more examples for two way anova which maybe useful for you :
http://www.sthda.com/english/wiki/two-way-anova-test-in-r
https://rcompanion.org/rcompanion/d_08.html
https://www.sheffield.ac.uk/polopoly_fs/1.536444!/file/MASH_2way_ANOVA_in_R.pdf
https://www.statmethods.net/stats/anova.html
I would prefer modelling the growth depending on the copper concentration, what shold follow some kind of inhibition curve (sigmoidal with negative slope), so that the IC50 (or a similar statistic) could be estimated for the strains and compared between them. But the small number of different concentrations is a limitation (a sigmoidal model would need at least 3 parameters: initial height, IC50 and slope at the IC50.
Using ANOVA is a poor solution. The ANOVA just tells you is a restriction in a model causes significant problems to describe the observed data. The only sensible question you can adress with ANOVA directly is if a model including the CU_CONC:TRICH interaction performs better than a model without that interaction:
model
Seyed Mahdi Amir Jahanshahi Thank you for your help.
Jochen Wilhelm Thank you for your advice. Unfortunately I have to work with what I've got. The experiment was planned this way, because I have interactions with 4 more metals. In the future I will consider your advice.
Yes, this is what goes wrong all so often. Statistics is needed to plan an experiment, not to analyze the data.
From R. A. Fisher, 1938:
To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of.
That's just what happened :)
To paraphrase Jochen Wilhelm , Wouldn't it be better to do 1 successful, publishable experiment rather than 4 that weren't?. Show these posts to your advisor and propose that the experiment be changed before everything has to be discarded with no usable result. Good luck, David Booth PhD(Phys Chem), PSTAT, Emeritus Prof and director of 10 Doctoral dissertations, and undergraduate doer of TMTC growth curves of N. crassa and S. cerevisiae in various mutagens under the direction of Prof. H. E. Brockman. PS Prof Wihelm is right about planning experiments. Consider the stats before you collect data.
David Eugene Booth Thank you for your advice. I will try to fix this as soon as I can get back to the lab.
Wiktoria Pyrkosz I found some references that may be of value to you. See attached protocols, Good Luck, David Booth
http://fgsc.net/neurosporaprotocols/How%20to%20use%20chemical%20mutagenesis.pdf
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA