Using t-test to find differences
As intellectual investigators, we want to determine if a difference observed between the mean values of 2 groups are due to something of interest rather than due to chance alone.
The formula for t is:
The top of the formula is the "signal" - i.e., the variance between groups
The bottom of the formula is the "noise" - i.e. the variance within groups
If there's a real difference between the groups, we expect the signal to be louder than the distracting noise :)
The bottom of the formula is the "noise" - i.e. the variance within groups
If there's a real difference between the groups, we expect the signal to be louder than the distracting noise :)
Unpaired Sample t-test
We use this when 2 groups are not related in any way.
This is also known as between-subjects t-test / unpaired-samples t-test.
This is also known as between-subjects t-test / unpaired-samples t-test.
Paired Sample t-test
We use this when we measure one group at two different times, to see if there's a difference in the means. For example, when we want to see if students in one class have improved on their math scores.
This is also known as within-subjects t-test / repeated-measures / dependent-samples.
This is also known as within-subjects t-test / repeated-measures / dependent-samples.
One Sample t-test
We use this when we want to compare one group with the larger population.