How do I interpret spline range estimates?

asgee — Tue, 10 Nov 2020 01:03:36 GMT

Hi all,

Might be a really dumb/simple question but I'm trying to remind myself of what the spline estimates represent when you run a PROC LOGISTIC (or any model with splines) for that matter.

I ran my model using the KNOTMETHOD=PERCENTILELIST(5 35 65 95), which gave me knots placed in these 4 points of my data:

Knot1 = 6

Knot2 = 17

Knot3 = 26

Knot4 = 59

However, when I look at my analysis of maximum likelihood estimates table, the first 4 lines show:

Intercept

Spl1

Spl2

Spl3

My question is, what's the range that these Spline Estimates represent? Does "Spl1" mean the estimate between 0-6? Or 6-17? If it does mean 6-17, why is it not reporting the estimate from 0-6 (and respectively the upper range of 59 or greater)? Or is there some reason that it is omitting those ranges (could be too extreme?).

Any help clarifying this would be very much appreciated!

Re: How do I interpret spline range estimates?

Rick_SAS — Tue, 10 Nov 2020 14:05:14 GMT

It is not a dumb question. For the interpretation of what each spline component means, see "Visualize a regression with splines."

topic How do I interpret spline range estimates? in SAS Programming

How do I interpret spline range estimates?

Re: How do I interpret spline range estimates?