# Which tests, within-subject experiment

#### GBM

##### New Member
Hi! I am testing the effect of exercise on memory, at 2 time points after the session (15/30 min). In-subject and control groups. What tests to analyse the effect and what for comparisons, please? Thanks for any help!

#### fourize

##### New Member
Hello,
If I understand correctly, the variable to be analyzed is continuous, right? If so, a repeated measurement ANOVA would be suitable.

F.

#### GBM

##### New Member
Hello,
If I understand correctly, the variable to be analyzed is continuous, right? If so, a repeated measurement ANOVA would be suitable.

F.
Thank you so much for your help! Dependent variable is Recollection (memory test). One measurement will be taken before the exercise and two measurements after the exercise - one at 15 and one at 30mins after. Independent variable = Exercise. Two groups: Exercise group and a control group (sedentary). The two groups are Categories of Independent variable? Sorry, a beginner!

#### Karabiner

##### TS Contributor
So you could perform a "mixed" analysis of variance, with a repetead-measures
factor "time of measurement", and a between-subjects factor "group". The interaction
between these factors will tell you whether change over time was different between
groups.

With kind regards

Karabiner

#### GBM

##### New Member
Using ANOVA or pair sample t-test? I thought two t-tests, for two time measurements would show the difference in reaction time before and after? I don't know which test to use for which outcome. And what test should I use for 'between-subject factor 'group'? In very simple language if possible...Thank you!!

#### Karabiner

##### TS Contributor
Seemingly, you are not accustomed to analysis of variance? There, the main effects of time,
group and the interaction will be tested within the model.

If your groups were created by randomization, you could consider to calculate difference
values (e.g. between baseline and follow-up 1), and compare them between groups using
a independent samples t-test.

With kind regards

Karabiner

#### GBM

##### New Member
Yes, I am a beginner and so far have learned t-tests and introduced to ANOVA. I am just not sure which stats test for which result to use. To test the difference in performance before and after the exercise, I can use paired-sample t-test? Then, compare the means of the two time-points? And for the difference between 'experiment' and 'control groups, is it independent t-test or ANOVA?
BTW: Which test would you use for variance? Just for future reference...

#### Karabiner

##### TS Contributor
To test the difference in performance before and after the exercise, I can use paired-sample t-test?
Yes, but you can only test whether the total group showed a statistically significant
change between time points, or, whether each group separately showed a statistically
significant change. But seemingly, you want to know whether the change from
baseline to follow-up was larger for the experimental group than for the control
group? The paired-samples t-test won't tell you that.

And for the difference between 'experiment' and 'control groups, is it independent
t-test
If groups were created by randomization, you could compare follow-up memory
between groups using an independent samples t.test, yes.

I suggested to first calculate the change from baseline to follow-up for each subject,
and then to compare these change values between groups using independent
samples t-test. Using the difference values would increase statistical power,
because you would use more statistical information (about the pre-existing differences
a baseline)

Which test would you use for variance? Just for future reference...
What do you mean?

With kind regards

Karabiner

Last edited: