Solved
Contributor
Posts: 46

# Generating Slope for regression line

Is there a way to get the slope coefficient for each group separately?

I used this code to generate the model and found a significant interaction effect. I wanted to generate some plots for these significant ones.

proc mixed data=&data namelen = 200;
class &group;
model &response = &exposure &group &exposure*group + confounders;
ods output tests3 = Fulltest&i;
run;

I have this code to try and generate the graphs:

ods graphics off;
proc reg data=&data;
model &response = &alcoholvar;
ods output ParameterEstimates=PE;
run;

data _null_;
set PE;
if _n_ = 1 then call symput('Int', put(estimate, BEST6.));
else call symput('Slope', put(estimate, BEST6.));
run;

PROC sGPLOT DATA = &data (where =( &group ne .)) noautolegend dattrmap = attrmap ;
reg x = &x y = &y /group = &group clm clmtransparency = 0.4 attrid = &group;
title "&title";
inset "Intercept = &Int" "Slope = &Slope" /
border title="Parameter Estimates" position=topleft;
run;

This only generates the intercept and slope for the first group listed. Is there any way to generate the slope for each group, (in my case there are 3 groups)? So there should be a slope for the line with group = 0, an 2nd slope for the line with group = 1, and a 3rd slope for the line with group = 2.

Accepted Solutions
Solution
‎01-09-2018 12:03 PM
Super User
Posts: 13,084

## Re: Generating Slope for regression line

Proc reg may generate the plots you want, or at least give you an idea of what the data looks like. But I suspect that you need to tell Proc Reg which variable is your group variable with a BY statement. Sort the data by the group variable before Proc Reg and add a By statement.

The parameter estimates will then have 3 values, one for each group and you will need to create 3 macro variables for intercept and slope to reference in the inset statement. Look at the values of the parameters data set.

All Replies
Solution
‎01-09-2018 12:03 PM
Super User
Posts: 13,084

## Re: Generating Slope for regression line

Proc reg may generate the plots you want, or at least give you an idea of what the data looks like. But I suspect that you need to tell Proc Reg which variable is your group variable with a BY statement. Sort the data by the group variable before Proc Reg and add a By statement.

The parameter estimates will then have 3 values, one for each group and you will need to create 3 macro variables for intercept and slope to reference in the inset statement. Look at the values of the parameters data set.

☑ This topic is solved.