Hi i am doing a physiology lab report graph concerning the effects of amiloride on na channels. We have more data values before amiloride was added compared to after and need to do a paired t test but excel states error. I assume this is due to unequal sample sizes. How would one go about this?
Regards, luke
Hey Luke.
A paired t-test when you have unequal sample sizes does not make any sense, conceptually or mathematically.
Conceptually, a paired t-test is good for when your "before" values have a lot of variance, relative to the difference between your before and after values. Moreover, it is basically only applicable when you have a "before" and "after" value recorded from a single "subject" (a subject could be a cell, a piece of tissue, or a human etc). Really, what it is asking is "is there a systematic difference between the before and after?"
Mathematically, a paired t-test is performed by first taking the difference between the before and after data. So you just end up with a big list of numbers. This big list of numbers is the subjected to a one sample t-test, to see if its mean is different from zero. i.e. if the difference is non zero, then there is a change between before and after.
Because of both of these ideas, you can see it makes no sense to perform a paired t-test if you don't have both a "before" and "After" for every single individual. If you don't have a before and after data point for a given subject, you need to remove it from your analysis, or.... perform a two-sample t-test.
Hey Luke.
A paired t-test when you have unequal sample sizes does not make any sense, conceptually or mathematically.
Conceptually, a paired t-test is good for when your "before" values have a lot of variance, relative to the difference between your before and after values. Moreover, it is basically only applicable when you have a "before" and "after" value recorded from a single "subject" (a subject could be a cell, a piece of tissue, or a human etc). Really, what it is asking is "is there a systematic difference between the before and after?"
Mathematically, a paired t-test is performed by first taking the difference between the before and after data. So you just end up with a big list of numbers. This big list of numbers is the subjected to a one sample t-test, to see if its mean is different from zero. i.e. if the difference is non zero, then there is a change between before and after.
Because of both of these ideas, you can see it makes no sense to perform a paired t-test if you don't have both a "before" and "After" for every single individual. If you don't have a before and after data point for a given subject, you need to remove it from your analysis, or.... perform a two-sample t-test.
I am in agreement with the previous answer. The ideal is to perform a paired t-test when you have values for before and after for the same sample. (exclude those that you do not have both values for). Alternatively you can perform a two-sample equal variance or two sample unequal variance.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology