Statistical analysis in agricultur and agronomy

#1
Hello all members,
I need your guidance about the analysis, as you are good at statistics.
I explain my experimental design.
Actually I inoculated 3 microorganisms (ASS-ASL and ASY using different concentrations i.e. 50-75 and 100ul/L). I wanted to check if these organisms at different concentrations affect the growth of wheat seedlings. I had 4 replicates for each treatment and 4 CK control for each treatment (No inoculation of microgranisms). As I am not clear, I can send you my raw data. I checked if different dosages of microbes promote or demote the biomass of wheat seedlings including (SHOOT LENGTH , ROOT LENGTH, FRESH WEIGHT, DRY WEIGHT SHOOT FRESH ROOT WEIGHT, ROOT DRY WEIGHT). I also think that variables may be correlated. Please note that each of the experiment is independent. Why I say variables correlated? The reason is below ground parameters may be related to above ground parameters (Shoot length, fresh weight, dry weight).
According to my understanding I want to do ANOVA and pot pretty graphs (2-3). Which test I should apply? one way Anova, two way anova or Manvoa. I also attach my data. Time is really out for me. I will highly appreciate the help. If some one is willing to help in detail, I can consider him/her to put in author list, as I am preparing manuscript.
Could you please help me in this matter?
Best wishes,
Abdul
 

Attachments

staassis

Active Member
#2
You do not use ANOVA here because the tested levels of the three predictors (ASS. ASL and ASY) are scale variables. The numbers actually have a meaning. They are not just labels for separate categories. E.g. the concentration of 50 is exactly half of the concentration of 100.

Use the three concentrations and, possibly, their non-linear transformations as predictors in a linear model. In this model, the dependent variable is Growth or some non-linear transformation of it. You can decide on the optimal non-linear transformations, or the lack of such, based on the goodness of fit of the estimated models.