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 in Proc Glimmix

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
- Highlight
- Email to a Friend
- Report Inappropriate Content

10-09-2009 06:05 PM

Hello- I have run a few models with glimmix and there is something that has confused me and still does. I hope someone can help me figure out what is going on. Right now, I am working on some data whose outcome variable is binary. There are also repeated measurements. So if I run a 'relatively' simple model, say:

proc glimmix data=subset5;

class wave hhid;

model injury(event='1')= arthrit male wave wave*arthrit wave*male / s d=b;

random _residual_/ sub=hhid type=cs;

run;

Where the outcome injury and the covariates arthrit male are coded as 1 or 0.

So I obtain my Type III tests of fixed effects. The p values for these are:

Num Den

Effect DF DF F Value Pr > F

arthrit 1 2974 33.73 <.0001

male 1 2974 16.45 <.0001

wave 3 1911 1.83 0.1387

arthrit*wave 3 2974 0.64 0.5902

male*wave 3 2974 0.35 0.7880

However, when I look at the solution, the p-values do not correspond at all. For instance, the p-values for simple effects such as arthrit and male are:

Standard

Effect wave Estimate Error DF t Value Pr > |t|

Intercept -2.0517 0.3033 905 -6.76 <.0001

arthrit 0.4077 0.2432 2974 1.68 0.0938

male 0.2071 0.2719 2974 0.76 0.4464

Can someone please explain to me what is going on? Thank you in advance.

Andres

proc glimmix data=subset5;

class wave hhid;

model injury(event='1')= arthrit male wave wave*arthrit wave*male / s d=b;

random _residual_/ sub=hhid type=cs;

run;

Where the outcome injury and the covariates arthrit male are coded as 1 or 0.

So I obtain my Type III tests of fixed effects. The p values for these are:

Num Den

Effect DF DF F Value Pr > F

arthrit 1 2974 33.73 <.0001

male 1 2974 16.45 <.0001

wave 3 1911 1.83 0.1387

arthrit*wave 3 2974 0.64 0.5902

male*wave 3 2974 0.35 0.7880

However, when I look at the solution, the p-values do not correspond at all. For instance, the p-values for simple effects such as arthrit and male are:

Standard

Effect wave Estimate Error DF t Value Pr > |t|

Intercept -2.0517 0.3033 905 -6.76 <.0001

arthrit 0.4077 0.2432 2974 1.68 0.0938

male 0.2071 0.2719 2974 0.76 0.4464

Can someone please explain to me what is going on? Thank you in advance.

Andres

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

10-14-2009 06:48 PM

Andres,

The p-values for male reported in the Parameter Estimates table represents a test of the effect of gender in the reference wave only. But the p-value for male reported in the Type III tests of fixed effects table represents a test of the gender effect across all waves.

To better understand this, try adding the following ESTIMATE statements to your GLIMMIX run:

estimate "Male effect in reference group"

male 1

male*wave 0 0 1;

estimate "Average male effect"

male 3

male*wave 1 1 1 / divisor=3;

The first ESTIMATE statement should give you the effect estimate and p-value returned in the Parameter Estimates table. The second ESTIMATE should give you the p-value returned in the Type III tests of fixed effects table.

Your interest is probably not to test the effect of gender in wave 3 only. Rather, you probably would like to test the effect of gender in the entire data set. That is the test given by the second ESTIMATE statement representing the "Average male effect" across all waves and also the test presented by the Type III tests of fixed effects table.

HTH,

Dale

The p-values for male reported in the Parameter Estimates table represents a test of the effect of gender in the reference wave only. But the p-value for male reported in the Type III tests of fixed effects table represents a test of the gender effect across all waves.

To better understand this, try adding the following ESTIMATE statements to your GLIMMIX run:

estimate "Male effect in reference group"

male 1

male*wave 0 0 1;

estimate "Average male effect"

male 3

male*wave 1 1 1 / divisor=3;

The first ESTIMATE statement should give you the effect estimate and p-value returned in the Parameter Estimates table. The second ESTIMATE should give you the p-value returned in the Type III tests of fixed effects table.

Your interest is probably not to test the effect of gender in wave 3 only. Rather, you probably would like to test the effect of gender in the entire data set. That is the test given by the second ESTIMATE statement representing the "Average male effect" across all waves and also the test presented by the Type III tests of fixed effects table.

HTH,

Dale

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

11-03-2009 03:34 PM

Thank you very much. It makes sense now.