I am using ANOVA and T-Test to compare wheat grain characteristics between states, agroecological zones, soils, etc.
Most of this data is grain mineral concentrations but some of it is ratios describing grain mineral bioavailability. For example, a phosphorus fraction expressed as a percentage of total phosphorus.
My question is, can I use compare means if the means are generated from ratios? Or do I have to transform the data first? I'm asking because my supervisor gave me this comment:
I believe the estimates of mineral bioavailability as a molar ratio is that a ratio. Data that involves ratios should be transformed accordingly for proper comparison. There is no reporting on data transformation in the statistical analysis.
