I am trying to find any existing work that use pairs of participants in a repeated measures design where the participants are combined with participants of the other pairs in order to reduce the number of participants required. I have not found any design carried like that so far. If anyone can give me a helping hand on a reference it will be very much appreciated.
Hi, Jose:
If you plan to use a repeated measure design, then it would usually be the same group of people you test with, hence the "repeated". Between these two or more repeated experiments, you would apply the so-called "treatment" to the same group. The treatment could be adding a newly developed technology aid in the second experiment or simply teaching them some related things in-between the experiments. Then you would collect the data (quantitative, I assume) and do a t -Test or Anova analysis to see if the same group would perform differently before and after your treatment.
As for the number of participants, you usually need more than 30~35 people in order to fit the normal distribution presumed.
Hope these help.
Cheng-Hung Lo
Thank you Cheng-Hung . The part I am worried about is how to run the pairing of participants. Let’s say I have 6 participants for a day of testing. Instead of running three experiments with them I want to mix them so I can run up to 11 experiments with the same 6 participants by mixing them. That is where the problem arises. I believe I can account for learning effect but I need a solid argument that it is feasible.
Hi, Garcia:
I suppose you have already considered using "Balanced Latin Square" to schedule your participants?
Cheng-Hung
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Human-Computer Interaction