This is one of the benefits of using PROC GLIMMIX, you can do a much better job at modelling your data correctly. Notice that we have a random effect of Block? With PROC REG we cannot account for any random effects that may be in our design. To run this please use the following coding: Now many of us are using PROC GLIMMIX and it is just one of the many PROCs in SAS that can also estimate a linear regression equation.
To review the output in PDF form, please download here.įrom our output our simple linear regression equation is: It gives us results (if successful) regardless. So let’s finish it off by adding the quit. Quit <- We need to add a Quit for PROC REG, it will keep running in the background until we give it more instructions. Model absorb = glucose This is our Y = mX + b – we only have the Y and X in our dataset – absorb = Y and glucose = X. Proc reg data=linear_reg Calls the PROCedure REG and using the linear_reg dataset However, we will concentrate on 2: PROC REG and PROC GLIMMIX.ĭownload SAS coding for Example 5.3 here. As statistical procedures mature and new SAS PROCs are developed, it turns out that you can run this analysis in several PROCs. We will be using Example 5.3 to demonstrate a Simple Linear Regression. No worries, it is fairly straightforward, as long as you remember your goal of Y = mX + b. Of course! we can do this! But, how in the world do we accomplish this in SAS. We also tend to remember what the different parts mean:ī = Y-intercept or where our line crosses the Y-axisĪs soon as we say Y=mX + b, the anxiety that accompanies statistics seems to disappear. Our math teachers in high school did a great job with this equation, since most of us remember it.
It’s funny how many people I talk with that remember the following phrase: A Hitchhiker’s Guide to Statistics in Biology: Generalized Linear Mixed Model Edition. I will be using examples from Bowley, S.R.
The last workshop for the S18 SAS series will deal with estimating linear curves using PROC REG and PROC GLIMMIX and estimating non-linear curves using PROC NLIN and PROC NLMIXED.