I have an experiment where there is a between subjects factor and a within-subjects factor. Assume that the between subjects factor is "Gander" with two levels - male and female. The within-subjects factor is "ExerciseType" with three levels - push-ups, squats, and crunches. The Dependent variable is "HeartRate".

Now assume that "N" male participants perform each exercise and their heart rates are recorded after each exercise, and similarly "M" female participants perform the same set of exercises and their heart rates are recorded. After the completion of the experiment, I remove recorded heart rate values that are above and below some threshold, because I know that those are due to an instrument error.

Hence, in each cell, I end up with some 'x' number of values, and the value of x changes for each combination of (Gender, ExerciseType). Please refer to the attached images.

At this point, if I conduct mixed model repeated measures ANOVA in SPSS ( Analyze -> General Linear Model -> Repeated Measures), I do not receive any error though I was expecting that it might complain about the missing values.

So, what is happening within SPSS? Also, is such an analysis valid (I am curious because there are missing values)? If not, what other tests I should do?

Thank you.

enter image description here enter image description here


What you estimate is a repeated measures analysis with imbalanced clusters due to the omitted measures. It is inefficient and prone to certain forms of bias relative to completing follow-up on all participants. Particularly problematic is that those with naturally high, low, or variable heart rates are more likely to be omitted by this design. The best thing to do is not delve into data analysis but find out how to better measure heart rate.


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.