At what point is more correct to pool data together?

ste.lu

New member
Joined
May 25, 2018
Messages
1
Dear,

My premise is I am not a mathematician but I have a question which I think could be formalized and answered in a mathematical way.

I have a 3 different pool of data which have to go through the same process. In the end I want to pool all of them together to analyze them. When is more correct to pool them together:
a. Pool the samples together and send them to the process as a bulk and then analyze them
b. Send the data into the process as individual entity and pool them together after that but before the final analysis

Thank you.
 
Dear,

My premise is I am not a mathematician but I have a question which I think could be formalized and answered in a mathematical way.

I have a 3 different pool of data which have to go through the same process. In the end I want to pool all of them together to analyze them. When is more correct to pool them together:
a. Pool the samples together and send them to the process as a bulk and then analyze them
b. Send the data into the process as individual entity and pool them together after that but before the final analysis

Thank you.

This depends on your pooling. If you have assumed the three samples have equal variance, you had better prove that before you throw them all together. That's just one example.

If one sample is VERY expensive, making you pick a very small sample size, then it is unlikely your data will be as useful without pooling. I once read a fascinating paper on, essentially, "Pooling Sometimes". It examined the validity of combining very small destructive sample tests with computer simulated results. This might be useful in testing aircraft, for example. You don' want to crash too many aircraft to see what they look like when they hit a mountain. Again, this is just one example of the kind of thinking you may need to apply to your task.

If you're not sure, you should consult a professional statistician.
 
Top