topic Re: How to get a pValue for 2 numeric columns in SAS Programming

How to get a pValue for 2 numeric columns

Sathish_jammy — Sat, 03 Nov 2018 06:29:14 GMT

Hi Community,

I want to know, which procedure is used to find the P value of 2 numeric columns.

Here i attch the sample dataset.

data have;
input ID1 ID2;
cards;
10 15
. 30
20 38
50 20
15 10
80 50
25 40
. 50
;
run;

OP should be like,

	ID1	ID2	PVAL
<20	% Value ID1	% Value ID2	?
20 to 30	% Value ID1	% Value ID2	?
>30	% Value ID1	% Value ID2	?

I evaluated the % value, But I'm looking for Pval

And the one more question is : What is the right format to calculate the pval., (with missing value or only with available value)

Please help me to solve this.

Thanks in advance!

Re: How to get a pValue for 2 numeric columns

FreelanceReinh — Sat, 03 Nov 2018 09:42:33 GMT

Hi @Sathish_jammy,

There is no such thing as the "P value of 2 numeric columns."

What statistical test do you want to perform? What are the null and alternative hypotheses?

Ideally, you can find this information in your statistical analysis plan.

Re: How to get a pValue for 2 numeric columns

Sathish_jammy — Sat, 03 Nov 2018 10:30:11 GMT

Dear @FreelanceReinh

Thanks for your suggesstion.

I want to generating a report like the image below.

Here the orange bar is HBA value(Baseline) and the blue bar is HBA Followup

The bar diagrams are performed as %, If u calculate the values it ends up with 100.

I need to evaluate the p value for baseline and followup.

Re: How to get a pValue for 2 numeric columns

PaigeMiller — Sat, 03 Nov 2018 12:25:10 GMT

Sorry @Sathish_jammy, this explains nothing, your data does not relate to this diagram, and furthermore, this diagram does not really explain what the p value is that shows on the plot.

Please explain the results you want, in relationship to the actual data you posted. Please explain, as requested by @FreelanceReinh, the exact statistical test you want to perform so we can know how to compute the p-value.

Re: How to get a pValue for 2 numeric columns

FreelanceReinh — Sat, 03 Nov 2018 13:33:45 GMT

Unlike your initial post, your graph suggests that you want to compare dependent samples (baseline vs. follow up) with ordinal level data in three categories. However, assuming that this is about HBA1c values (i.e. metric values) in diabetic patients (?), it's not clear why you would categorize them rather than compute differences (i.e. change from baseline) and use these as input for a statistical test (e.g. a paired t-test or Wilcoxon signed-rank test). Even if the categorization made sense: Why would you perform three tests (McNemar tests?), one for each of the three categories, although they are not independent?