I have a datasets of populations with a response. I triplicated the experiment, so I have 3 trials of data, but the response rate likely varies between trials. It is a hypoxia response rate, which is sensitive to things like humidity/temperature that I can't perfectly control.
Each trial consists of a population with different treatments, plus empty vector controls.
So like this, but with like 50 entries for each treatment/control per trial:
Trial Response
1 TreatmentX 999
1 Control 576
2 TreatmentX 1111
2 Control 700
3 TreatmentX 700
3 Control 350
So for this example the treatment is always above control, but control varies alot. I was thinking of using Dunnetts test to see which treatments are different from control, but I want to normalize the data to account for the trial day differences. The population sizes vary so I can't just use the total average for each day, but I should be able to use my controls to normalize each trial. How do I do this in JMP?
If it's just about calculating the ratio (or difference) per trial , you can use the Normalization AddIn from https://marketplace.jmp.com/appdetails/Normalization+GUI
e.g. with the settings:
- column to normalize : Response
- function: ratio or difference
- grouping: by trial
- subset: Control
[The default aggregation is median, which works well even with just a single control and is much faster than the other modes.]
