Linear Regression - Repeat Measurements

Hi Everyone,

I am fairly new to Matlab and have a problem using Linear Regression.

I have an experiment where I have taken multiple repeat measurements. To simplify my problem I have a 10 by 1 vector of predictor variables (independent variables) and a 10 by 5 matrix of response variables (dependent variables). I know that these data should follow a linear trend and I want to evaluate how linear they actually are.

I want to apply a robust least square fitting to the average of my dependent variables at each independent variable. I can do this using robustfit. My problem arises as I also would like to weight the least square fitting method by the variance of the dependent variables at each data point. I can't figure out how to do this.

Any help would be greatly appreciated!

Best,

Siri

Answer 1

[Star Strider]

Repeated measures problems have their own statistical requirements. I agree with Tom Lane’s answer to: Setting weights for robustfit command because I can’t improve on it. See the documentation for glmfit for details.

[Star Strider]

After looking around the Statistics Toolbox functions for a few minutes, I suggest you see the documentation: Set Up Multivariate Regression Problems. It seems to be closest to what you need.

(I apologise for the delay ... GMT-6 here.)

Answer 2

[Anna]

Hi Star Strider,

Thanks for your answer. I have had a look through the documentation for glmfit but I am still a bit confused how to use glmfit with multiple response variables (I assume that's how I would treat my repeat measurements?). Would you perhaps be able to give me an example code line? Do I need to use another function?

thanks,

Siri