# How to do a t-test with repeated measurements from 3 different devices?

9 views (last 30 days)
Oliver Chen on 23 Apr 2015
Answered: Udit Gupta on 23 Apr 2015
Hi, I have a question regarding t test. For example, I have a vector X with 100 measurements representing the accuracy of a medical device. 30 of them are repeated measurements from device 1, 30 of them are repeated measurements form device 2, and the remaining 40 are repeated measurements from device 3. The average value of the 100 measurements is 10%. In order to test the hypothesis H0: mu = 12%; Ha: mu~=12%, how should I conduct t test in MATLAB? Because of the within-subject correlation, I don't think I can simply do ttest(X, 0.12)? Your help is much appreciated.
Best, Oliver

Star Strider on 23 Apr 2015
See the documentation for the RepeatedMeasuresModel class.

Udit Gupta on 23 Apr 2015
You can do it in three steps.
H1=ttest(X(1:30),12);
H2=ttest(X(31,60),12);
H3=ttest(X(61,100),12);
If you want to determine if the three deceives themselves have statistically significant differences or not you can use.
H=ttest(X(1:30),X(31:60)); % and so on.
Is this what you were looking for? If not please elaborate the question.