Questions about a Growth Model Calculation I want to perform

#1
Dear Talk stats members and fellow statistics lover,

I got Data from an Experiment with four Conditions of different Treatment and I want to know if the HRV improved significantly over the course of the therapy.

In all the examples I saw before, the Variable Time & Time_numeric always resembled the same timepoints for all participants.
In my Experiment it was not possible to keep the interval between the sessions identical, due to different workschedules of the participants, so I'm asking myself if it causes a problem?

Instead of the classic book example I saw for a time variable with 0, 3 , 6 months as datapoint for all participants my data looks more like: Patient x: 0,4,10,12.... (measured in Days) Patient y: 0,3,6,13....
 

katxt

Active Member
#2
There are things you can try but how valid and useful they are depend on the data and the situation.
- You could try fixing some time points and interpolating the data when you don't have the exact day. Fix a timepoint say 5 days. If it's there fine, otherwise work out your best estimate from, say, Day 4 and Day 7.
- Average the improvements by regressing HRV on days and allocating each subject that one number.
- Buy an hour of your local statistician's time to discuss the problem. A small price to pay in a project that looks as if it could have serious medical implications. This is my recommendation.
 
#3
There are things you can try but how valid and useful they are depend on the data and the situation.
- You could try fixing some time points and interpolating the data when you don't have the exact day. Fix a timepoint say 5 days. If it's there fine, otherwise work out your best estimate from, say, Day 4 and Day 7.
- Average the improvements by regressing HRV on days and allocating each subject that one number.
- Buy an hour of your local statistician's time to discuss the problem. A small price to pay in a project that looks as if it could have serious medical implications. This is my recommendation.
Thank you very much katxt!
I considered talking to a statistician already and it yielded a good result :)