# Repeated Measure ANOVA with muliple samples per participant

#### blooop

##### New Member
I am currently doing an experiment where I measure the time it takes for a participant to complete a task. There are 7 different modes for the task and I test each participant in every mode. I randomize the order that each participant does the modes.

This is the format of the data I'm recording. In the real experiment there are 7 modes (instead of 3 here) and each participant attempts the mode 10 timeS (instead of 3 here)
Code:
Participant ID	Attempt number	Mode 1 time	Mode 2 time	Mode 3 time
0	               1	   8.3	         6.9		        2.6
0	               2	   9.1	         5.1		        5.2
0	               3	   4.7	         7.6		  	10.0
1	               1	   2.3	         7.1		  	5.5
1	               2	   3.0	         9.7		  	8.2
1	               3	   3.3	         5.9		  	0.8
2	               1	   8.5	         2.1		  	0.8
2	               2	   2.5	         1.7		  	2.9
2	               3	   7.1	         8.2		  	8.0
I want to know which mode is the best, and if there is a statistically significant difference in the mean times between modes. From what I have read I think repeated measures ANOVA is the correct statistical test.

Last year I did another experiment with a very similar format and successfully used repeated measures ANOVA using SPSS. There was one thing that bothered me however. When inputting the data into SPSS, it did not seem to support inputting the raw data in directly. I had to take the mean of each participant's trial time, and I was not able to input the variance of their 10 samples into the ANOVA. It feels like I am losing useful data by taking the mean of all their trial times.

When I asked people about it, they said that this was due to the repeated measures experiment design. The difference between participants was more important than the difference in performance within each participant.

Question: Is it possible to use all 10 measurements per participant or is taking the mean of all their times and ignoring the variance of those times the correct way to analyse this?

#### Karabiner

##### TS Contributor
So you have 70 measurements (variables) and 2 repeated-measures factors, mode (7 levels), and attempt number (10 levels). You can use both factors, i.e. no need to aggregate, AFAICS.

With kind regards

K.