I have an experiment that includes 8 subjects during one treatment, measuring response variable. The hypothesis is that there is some correlation between AV and lactate during the treatment. Some subjects have a very late response and some earlier, it maybe depends on different physiology.

My idea was just to use the trend changes point from the fixed factor, In the figure some subjects have the trend change at time point 6, some have trend changes start at time point 10 leading to few data points for some and more for others. So, the data input to the mixed model is only the trend changes in data for every subject.

The fixed factor, because of the treatment, is changing all the time steps during the 2 hours experiment as does the response variable (AV).

So, if I look at the data plot with just linear modelling they have different slopes and intercepts for all subjects. I was thinking to model with random effects for intercept and the slope, see code below.


I'm a newbie, but I don't see any reason to drop the earlier time points. Try random slopes and intercepts and a quadratic term.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.