BookmarkSubscribeRSS Feed
☑ This topic is solved. Need further help from the community? Please sign in and ask a new question.
Calcite | Level 5

## How to Interpret Fit Diagnostic Plots for Proc Panel

Hi Everyone,

I am a new user, currently completing panel regression analysis for a university paper. The data I am using is a sample taken from the US Panel Study of Income Dynamics (PSID), with 23 variables and ~220k observations.

When I run a simple panel model, such as below, I do not get the observation plots, and the colored set of residual plots seem incomprehensible. Is it simply a case of so many observations making the plots over-complicated? Are these plots "normal" for proc panel, and if so how do I interpret them?

Thank you for any advice.

proc panel data=data plots=ALL;
id year id;
model ARSINH_wealth = pd age male white black hispanic otherrace education income employed divorce
marriage childbirth familydeath laidoff missedwork studentloan collegedegree ARSINH_socioeconomic / pooled;
title 'Model One - Pooled (OLS) Model (negative)';
run;

This question was originally posted to 'New Users' but I was advised to post it here instead.

1 ACCEPTED SOLUTION

Accepted Solutions
Rhodochrosite | Level 12

## Re: How to Interpret Fit Diagnostic Plots for Proc Panel

Perhaps the visuals get more meaning if you work with a sample of the data.

You could try subsetting your data.

``````data want;
set data;
sample_filter=rand("integer", 1, 20);
run;

proc panel data=want plots=ALL;
where sample_filter = 1;   /* in this case you would work with a sample rate of 5% */
id year id;
model wealth = pd age male white black hispanic otherrace education income employed divorce
marriage childbirth familydeath laidoff missedwork studentloan collegedegree ARSINH_socioeconomic / pooled;
title 'Model One - Pooled (OLS) Model (negative)';
run;``````

6 REPLIES 6
Calcite | Level 5

## How to Interpret Fit Diagnostic Plots for Proc Panel

Hi Everyone,

I am a new user, currently completing panel regression analysis for a university paper. The data I am using is a sample taken from the US Panel Study of Income Dynamics (PSID), with 23 variables and ~220k observations.

When I run a simple panel model, such as below, I do not get the observation plots, and the colored set of residual plots seem incomprehensible. Is it simply a case of so many observations making the plots over-complicated? Are these plots "normal" for proc panel, and if so how do I interpret them?

Thank you for any advice.

proc panel data=data plots=ALL;
id year id;
model wealth = pd age male white black hispanic otherrace education income employed divorce
marriage childbirth familydeath laidoff missedwork studentloan collegedegree ARSINH_socioeconomic / pooled;
title 'Model One - Pooled (OLS) Model (negative)';
run;

Super User

## Re: How to Interpret Fit Diagnostic Plots for Proc Panel

Better post it at Forecasting Forum, since proc panel is under SAS/ETS .

https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/bd-p/forecasting_econometrics
Calcite | Level 5

Ok, thanks.

Super User

## Re: How to Interpret Fit Diagnostic Plots for Proc Panel

When you need a post moved, call out to a Super User. We can do that for you.

I merged your posts.

SAS Super FREQ

## Re: How to Interpret Fit Diagnostic Plots for Proc Panel

Hello @jim_wong ,

@SASCom1 can help you with this.

Thanks,

Koen

Rhodochrosite | Level 12

## Re: How to Interpret Fit Diagnostic Plots for Proc Panel

Perhaps the visuals get more meaning if you work with a sample of the data.

You could try subsetting your data.

``````data want;
set data;
sample_filter=rand("integer", 1, 20);
run;

proc panel data=want plots=ALL;
where sample_filter = 1;   /* in this case you would work with a sample rate of 5% */
id year id;
model wealth = pd age male white black hispanic otherrace education income employed divorce
marriage childbirth familydeath laidoff missedwork studentloan collegedegree ARSINH_socioeconomic / pooled;
title 'Model One - Pooled (OLS) Model (negative)';
run;``````

Discussion stats
• 6 replies
• 3871 views
• 1 like
• 5 in conversation