topic P-value in SAS Procedures

P-value

robertrao — Mon, 03 Dec 2012 08:21:27 GMT

Hi Team,

I am comparing two means and calculating Z score and then the p-value.

If i get the p-value less than 0.05 will I call it "YES THERE IS SIGNIFICANT DIFFERENCE"

NO, THERE IS NO SIGNIFICANT DIFFERENCE???

Please give me some idea.

Regards

Re: P-value

Ksharp — Mon, 03 Dec 2012 08:52:09 GMT

What is your SAS code ?

If i get the p-value less than 0.05 , I can call it "YES THERE IS SIGNIFICANT DIFFERENCE" .

But I will more trust confidence interval .

Re: P-value

ballardw — Mon, 03 Dec 2012 16:04:44 GMT

You should decide what would be a significant difference BEFORE the test and know why you chose that value.

Re: P-value

rnmishra — Mon, 03 Dec 2012 16:24:17 GMT

The P-value =0.05, is not always useful. First go back to your problem and decide how accurate you want to be. Then decide on your p-value- not necessarily 0.05.

Then go ahead with what you are doing.

These are just my thoughts.

Thanks.

Re: P-value

Doc_Duke — Mon, 03 Dec 2012 17:52:03 GMT

There is a lot more to interpreting statistical significance than just looking at the p-value. See, for instance,

http://ceaccp.oxfordjournals.org/content/7/6/208.full

Your second claim, "NO, THERE IS NO SIGNIFICANT DIFFERENCE???" is a statement about the type II error and not about the type I (p-value) error. The correct interpretation statement is

"We failed to determine statistical significance.", which is much weaker.

Doc Muhlbaier

Duke

Re: P-value

SteveDenham — Tue, 04 Dec 2012 13:17:16 GMT

Great reference, Doc. I want to send it to all of the toxicology study directors

Steve Denham