I have two compounds with different concentrations in my project. For example, compound A has concentration of 0, 50, 100 and 200 uM and compound B also has concentration of 0, 50, 100 and 200 uM. In my case, I want to know which compounds with the particular concentration is the best for further research. What kind of statistical analysis can be used to compare the means between each concentration? can I use one way ANOVA?
It is not a combination. The samples were treated with chemical A and B in different experiment. They are independent events. I had done the analysis for each compounds and found the best concentration for that compounds. But I want to find out the best chemical with the best concentration. For example, the best results was given by 100uM of chemical A after I had done the analysis by comparing each concentration of each chemical (0uM A with 0uM B, 0uM A with 50uM B, 0uM with 100uM and so on). Is it possible by using analysis? someone suggested me to use two way ANOVA.
One-way ANOVA in your study will increase the false positivity rate. If the conditions of experiment at each conc. were similar then better you go for REPEATED MEASURE One-Way ANOVA.
In case you have observed some growth or suppression and your data is in the form of percentage (%), better you calculate ED50 (the medina effective dose). The smaller the value of ED50 the better is the compound.
You should define what you mean by "best for further research". Highest or lowest response values? Small or large variance (between chemicals and/or within chemicals)? Anything else? Combinations of that?
If your question is simple (e.g. "best = highest response") then just select the chemical and concentration for what you measured the higest average response. This is it. If you look for combined parameters, like "best = high response but also a low within-variance", you must somehow weigh the relative importance of these parameters and combine them into a single statistic that can be compared between the groups.
In either case, there is no need to perform any statistical test (you don't anyway have any clearly defined, resonable or sensible null hypothesis to be tested). If differences do not come out clear then different things like feasability, availability, toxicity, prices,... of the chemicals are much more relevant for your decision.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics