i'm fairly new to the field of statistics and i have an understanding problem when conducting an experiment.
I want to compare two groups/treatments and the experiment lasts for ten rounds. The participants perform on a task. I have two dependent variables (1) performance (metric) and (2) cheating on the task (binary: yes/no). I want to analyze if there is a significant difference in cheating behavior between the two treatments/groups.
To my understanding, i could collapse the cheating behavior across all ten rounds into one score and make mean comparisons, or i could use a general linear mixed model (with round as a within-subject factor).
However, my hypothesis is based on participants' cheating behavior ONLY when their performance is lower than a certain threshold. This means I am only interested in the fact whether or not the participants cheated if their behavior is below that threshold. If it is above this performance threshold, i am not interested.
Question: Can i use a general linear mixed model and only include each participants' data points for a certain round if his/her performance is below a certain threshold? Am I that flexible when using a regression analyis? This means that while some participants would contribute 10 data points (i.e. their performance is lower than the threshold in each of the ten rounds), while others for example only contribute 2 data points (i.e. their performance is lower than the threshold in two of the ten rounds and higher in eight rounds).
Do regression analyses allow that much flexibility in terms of what data to include when i am interested in answering such a narrowed hypothesis?