Hello everyone,
I have 25 participants, spread upon eight treatment conditions. I know so far that I will need to run a repeated measures ANOVA, since all 25P have been exposed to the aforementioned eight-conditions. I want to measure their performance for each condition, in this case completion time.
Here is where I get confused, each treatment condition consists of 18 tasks, which are all the same for each condition. The only thing that "changes" in the study are the conditions themselves.
So that gives me, 18 in total completion time observations for EACH participant per condition.
How would I conduct a RM-ANOVA in this case ?
a) Would I average out the 18 completion time observations that I have for each participant, effectively having 25 completion time observations per conditions ?
b) Would I consider the 18 completion time observations independent, thus 18 x 25 = 450 observations per condition ?
c) Would I average out the completion time of the 25 participants across the 18 tasks for each conditions having effectively 18 in total observations ?
The research question I'm trying to answer is whether each treatment (8 conditions), makes a difference in the completion time for each participant (25 participants), under varying task complexity (18 tasks with varying difficulty).
Your question is very complicated because my initial guess is your sample size is too small. First to check on sample size related question you want the program, G*power. See this link:
https://www.google.com/search?q=g*power&rlz=1C1CHBF_enUS847US847&oq=G*power&aqs=chrome.0.0l6.10286j0j8&sourceid=chrome&ie=UTF-8
Repeated measures is a special case of what are called longitudinal models, the best info that I know of on that is here:
https://www4.stat.ncsu.edu/~davidian/st732/notes.html
These are the best sources I know of to deal with your questions. Best wishes, David Booth
David Eugene Booth Thank you for your reply. I have analysed my data and there doesn't appear to be any outliers in the boxplot output I have written in python. In addition, from running a histogram, the distribution of the data seems also to appear Gaussian. I will take a look at the G*power program you recommended.
Hi. As I said I doubt 30 is a big enough sample size to meet your requirements as stated. Further you probably need a longitudinal design to handle your data set. Standard repeated measures ANOVA may not do it. Anyway the two links I gave you, should help. The first thing you want to do is see how big the sample should be to meet your requirements - G*power.. There is nothing magic about 30. Good luck, David Booth
BTW I think none of your 3 ways of handling the DV info is a good idea. I think you are in the world of multivariate analysis. This is exactly why you should consider the analysis step before collecting data. Good luck, David Booth
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA