# Statistical Test to Compare Two Methods

#### bpyoumans

##### New Member
Hello,

I sequenced some samples using two different sequencing methods and I want to compare the results. The two methods give very different amounts of sequence data so my outcome variable is relative abundance of bacteria at the taxonomic level of Phylum. I don't think McNemar's Test would be appropriate because my outcomes aren't nominal but maybe I'm wrong. Is there another test I can do (preferably in R)? I attached a sample of the data. I appreciate any help anyone can offer

E

#### elnaz

##### Guest
yes, you are right beacuse your data aren't nominal so you can't use McNemar's Test.
Can you tell me that for example your first sample with first method is from same thing that you took first sample with second method?
because we must know that your samples with 2 methods are independent or dependent.
If this question be clear, so we know how to compare two methods.

#### bpyoumans

##### New Member
I took one sample and split it in half. The first half was sequenced by one method and the other half was sequenced by another method.

E

#### elnaz

##### Guest
Actually i am confused with your sample what is this sample, your sample is biological sample that's a reason i wana know about it.
I think your samples are dependent, my reason is that becouse you took one sample from a special thing and you split it in half so source of both half sample is same thing
so i think your samples are dependent so the test that you must use is t- test with dependent samples.
But if you think that i understand your samples wrong and with my explanation you found your samples indepndent so use t-test with independent samples.
But dn't forget that befor doing t-test u must check conditions of it.
If it's condition didn't exist then u must use nonparametric test that is equivalent of t-test.

#### bpyoumans

##### New Member
I'm looking at bacteria in a stool sample . I extracted DNA and then split the DNA sample in half. I think you're right. They are dependent samples and I'll do a parametric/non-parametric t-test after checking the appropriate conditions. Thank you so much!!