Hi all,
I have a residual plot which is shown below and I am not sure if the variance is constant or not because I can see a funnel shape.
Could you please suggest me a sas code for a test to make sure that the variance constant or not constant
@novinosrin wrote:
Sir, requesting your clarification. Should there be constant variance, does it mean outliers are not present? how does it relate to influential point? I am seeking insight in the interpretation. Can you pen down a point or two plz
You can have outliers with constant variance.
I am looking for a more severe pattern before I decide that the variance is not constant. I can draw such a pattern on piece of paper, but that doesn't help here. But image the same residual plot, where the data points on the plot fit into a triangle, narrow at one end and wide at the opposite end, that would indicate to me that you have nonconstant variance. The plot shown, the data mostly fits into a rectangle, hence my opinion that there is constant variance.
Now you could have low variance in the middle and high variance at the edges, then the points on the plot might fit into two different triangles pushed together with the narrowest parts of the triangles in the middle and wide parts at the edges, but I don't see that either.
And there could be other patterns as well. That's why we use plots, because the human eye is the best pattern detector there is (although there are sitautions where it can be misled).
Visually, I am not seeing a problem. If the variance was not constant, you would see a much more severe and well-defined pattern.
I don't know if there is a formal test that takes residuals to determine if they have constant variance.
Sir, requesting your clarification. Should there be constant variance, does it mean outliers are not present? how does it relate to influential point? I am seeking insight in the interpretation. Can you pen down a point or two plz
I like your question. Sounds like very similar to the course I took. But we have a big boss @PaigeMiller who can help us both in our understanding. Let's wait
@novinosrin wrote:
Sir, requesting your clarification. Should there be constant variance, does it mean outliers are not present? how does it relate to influential point? I am seeking insight in the interpretation. Can you pen down a point or two plz
You can have outliers with constant variance.
I am looking for a more severe pattern before I decide that the variance is not constant. I can draw such a pattern on piece of paper, but that doesn't help here. But image the same residual plot, where the data points on the plot fit into a triangle, narrow at one end and wide at the opposite end, that would indicate to me that you have nonconstant variance. The plot shown, the data mostly fits into a rectangle, hence my opinion that there is constant variance.
Now you could have low variance in the middle and high variance at the edges, then the points on the plot might fit into two different triangles pushed together with the narrowest parts of the triangles in the middle and wide parts at the edges, but I don't see that either.
And there could be other patterns as well. That's why we use plots, because the human eye is the best pattern detector there is (although there are sitautions where it can be misled).
Merci! Thank you very much indeed!
SAS Innovate 2025 is scheduled for May 6-9 in Orlando, FL. Sign up to be first to learn about the agenda and registration!
SAS' Charu Shankar shares her PROC SQL expertise by showing you how to master the WHERE clause using real winter weather data.
Find more tutorials on the SAS Users YouTube channel.