# Which statistical test should I use?

#### stex3986

##### New Member
Hi everyone,
maybe a statistical expert could help me out.
I am interested in an event, that occurs in mean 30 times per month with an SD of 10. This relies on a data collected over a period of 1 year.
Now I want to compare if this event happens as often after the coronavirus pandemic.
So i have got 12 months of data compared to the 1 month now.
Which statistical test could i perform?

#### noetsi

##### Fortran must die
A question you should ask first is, can you generalize after 1 month. And whether 12 months of data is enough to generalize as well.

You might want to consider a paired t test if you think you can generalize (Which I doubt you can).

#### stex3986

##### New Member
Thank you very much. So it is that easy ok.
You are absolutely right, but it should be more of an hypothesis-generating work than actual statistical proof.

#### stex3986

##### New Member
A question you should ask first is, can you generalize after 1 month. And whether 12 months of data is enough to generalize as well.

You might want to consider a paired t test if you think you can generalize (Which I doubt you can).
I thought about the paired t-test. The problem is I would have unequal sample sizes - namely the data of the 12 months and the data of the 1 month.
Wouldn't that be rather independent?

#### noetsi

##### Fortran must die
Nothing I have ever read suggests that unequal sample designs have anything to do with whether a test is paired or independent. You are comparing means not individual cases. In some cases in ANOVA unequal designs create problems (when you lose cases I think) but that has nothing to do with independence.