Hypothesis Test Using PMCC Flashcards
Wait what is a SAMPLE PMCC, where did this come from?
When using data to work out a PMCC often it came from a LARGE parent data set, that has been SAMPLED from
Thus any conclusions from your data my not NECESSARILY mean the same thing for the real parent
Why may sample not be like parent
Because parent could be negative and your sample is positive, so it won’t work
So what’s the hypothesis test essentially we doing?
We calculating PMCC from sample, and DETERMINING if this is enough to CONVINCE us the trend is the same for thr parent, this is thr sample PMCC hypothesis test we trying to do
When they mean sample from parent give a context scenario
Just like you csn’t measure all fish in sea, yiu measure a bit, find a correlation but now you want to see if this will be same for everythign rlse
What is the significance level again DEFINITION
The probaiolty of rejecting the null hypothesis when it is ACTUALLY TRUE
Thus the lower the significance level, the better reliability, but less chance of a positive result
So need to balance significance level , but exam question might ask to discuss
What was 1 tail vs 2 tail again
1 tail is one direction only, positbe or negative correlation
2 tail is if there is some correaltion or any
Remember what To do with PMCC at the end to compare with fro tail value
Find the MOD. Of it. If it’s bigger, reject H0!
HOW TO DO CONCLUSIVE STATEMENT
Don’t go and make new statement, if you reject, then say there is INSUFFIECNET EVIDECEN TO SUGGEST whatever tedt was for
And if it is accept, say sufficient evidence to suggest there is whatever test saud for
What if they give you a P VALUE, and all you have is significance what to do
IMPORTANT
Carry out hypothesis test as normal
- if it is a 2 tail test, HALF SIGNFINICANVE VALUE, if not keep same
- NOW COMPARE P TO SIG VALUE
If P value is LESS THAN SIGNIFICANCE, reject
What’s the problem we face with LARGE samples from a parent and thus why do we use effect size
Large samples, the critical value becomes very small, and so it’s easy ti by chance find evdiecen to reject h0, when we shouldn’t
Thus effect size vs,use allow us to interpret PMCC for large data sets
So if I calculate large sample of data PMCC and it’s 0.5 what tries it mean
Using effect size graph, it means it’s a large effect, so string correeltiwon, even tho before wr said 0.8 and Above was STROng