Normalizing Data to Average of a Particular Condition
Feb 23, 2009 11:12 AM(2443 views)
Hi. I have a dataset that I would to normalize, and I would like to know the best way to go about it. The data table is structured (for our purposes) as three columns: subject number, experimental condition, and reaction time. There are five different experimental conditions. I would like to normalize the reaction time for conditions 2, 3, 4, and 5 to the average reaction time for condition 1 for each subject. This would essentially express each reaction time in units of each individual subject's reaction time to allow me to look at the impairment of reaction time for each individual as a function of the experimental condition.
So, I would like a formula or procedure to compute the average reaction time for experimental condition 1 for each subject and divide the reaction times for conditions 2-5 by the average for condition 1.
"I would like to normalize the reaction time for conditions 2, 3, 4, and 5 to the average reaction time for condition 1 for each subject."
This is unclear. Is there more than one observation for each subject at each condition? You didn't say that there were more than one observation for each subject at each condition.
It would be relatively easy to set each condition to have a mean of zero, which should also then give you the ability to do the comparison that I think you want to do. It is more complicated and time consuming to match everything to condition 1. Would zero means work for you?
Sorry for my lack of clarity. Yes, there is (are?) more than one observation for each subject at each condition. I think I would prefer to divide the responses by the average response for condition 1 rather than zero the means. In the meantime (no pun intended :-), how would I go about zeroing the mean for condition 1 for each subject?
If you need to normalize via the means of condition 1, that seems to require as series of steps in JMP to achieve this.
Here is an outline of the steps
1. Use Tables->Summary to obtain a table of means by subject for condition 1 2. Join this new table back into the original table, by subject 3. Create a new column where you create a formula to do the normalization
To get the same data to have zero means for each subject/condition combination, you would do a Fit Model, where the term in the model is subject*condition. The residuals have zero mean in each subject condition combination.