turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- P values for proc logistic explanatory variables i...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

11-30-2016 11:15 PM

Hi,

Using v9.4.

Am trying to get p-values for each of my explanatory variables from a logistic regression using proc logistic.

Code looks like this, and the outest file looks fine, except I'd like to see individual p values in the output.

proc logistic data = regc.r_sec_&i. noprint outest=creg.m2_s&i.;

model y (event = '1') = duration lvol1 lvol2 lvol3 ispread1 mkt1 mkt2 mkt3 c_dist1 dist_x_m connect connect_x_dist orderimbal;

run;

Any help would be great - thanks.

Ed

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to tsetredst

11-30-2016 11:23 PM

What do you mean by individual p-values?

Have you looked at the Parameter Estimate table, after removing noprint?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Reeza

11-30-2016 11:46 PM

Hi Reeza,

Thanks for your response.

I am hoping to find the significance of each of the explanatory variables.

When I print the output, I get a table 'Analysis of Maximum Likelihood Estimates' that gives me the estimate, error, wald-chi-square and Pr>ChiSq.

I guess I have two questions form that:

1. is there a way of outputting those values for each of the input parameters into a separate table? (without having to open the results veiwer each time, as I am running a few regressions).

2. what can I infer from the Pr>ChiSq value? is it similar to a p-value in the sense that a small ("<.0001") is highy significant?

I am doing further reading on Q2 - as this is obviously more fundamental to the research I am conducting.

Thanks,

Ed

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to tsetredst

12-01-2016 12:12 AM

Add a line to your regression, it will create a data set called PE with the desired output.

ods output ParameterEstimates=PE;

2. That is the p-value. I HIGHLY recommend reading up on this as well as P-Hacking.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Reeza

12-01-2016 12:22 AM

Thanks Reeza.

That worked well.

And yes, thanks for that advice, noted and taken.

Ed