Paired-T or Independent T?

#1
Good evening,

Suppose I have 2 datasets, one for ETF gold prices(Gold ETFs are units representing physical gold) and one for the price of physical gold.

The sample size is the same for both data sets.

The samples are grouped according to their date, that is to say a price in physical gold has a corresponding price in ETF gold.

My question is, if I were to run a hypothesis to find out if there is a significant difference in their means, which test should I perform? The Paired-t or independent t?

Thank you
 

Miner

TS Contributor
#2
It depends on whether the samples are dependent or independent of each other. If there is some form of dependency between the samples use a paired t-test. If they are independent of each other use a 2-sample t-test. In your case the only way that I can see a possible dependency is if on each day, you took a sample of gold ETFs and a corresponding price. A quick and dirty test would be to create a scatterplot between your "pairs". If there appears to be a relationship between the pairs, you can probably used the paired t-test. No relationship, use the 2-sample t-test.