Working Plots for Bivariate Data
It is vital that you be able to make bivariate data plots. You will use them for a lot of purposes. Some plots will show your raw data as a scattergram and others will show the trend lines that best portray the relationship between your variables. Other plots are used for diagnostic purposes, such as showing the pattern by which your data vary from the trend line; these are "residual" plots.
The SAS analysis procedures, like PROC GLM, do no plotting themselves. Instead, they let you output values to a data matrix that can be used in the plotting procedure.
The SAS plotting procedure is PROC PLOT. It is really quite comprehensive and gives you considerable control over the scaling and appearance of your plot. The discussion here will avoid all but the most important capabilities of this procedure. The reason is that no matter how functional the program and how hard you try, plotting with a coarse printer (like a typewriter) will never show the sorts of detail that is necessary. Therefore, use PROC PLOT as a tool to get rough working plots. There are good alternative ways (such as with a number of personal computer packages) to get higher quality plots.
Note carefully that the purpose of PROC PLOT is to establish a cartesian-coordinate system, scale and draw the axis lines, and label these axes with variable names (or the value of a label if one has been associated with the variable name). It puts letter, numbers or symbols inside the plot to indicated the values in a data matrix.
What it doesn't do is an analysis of your data. It will not fit curves through your data values (that is done in PROC GLM, for example).
As a result, you will almost always use PROC GLM and PROC PLOT together. One does the analysis while the other plots the results. 168