Example 38.15 Comparing Multiple B-Splines

This example uses simulated data to demonstrate the use of the nonpositional syntax (see the section Positional and Nonpositional Syntax for Contrast Coefficients for details) in combination with the experimental EFFECT statement to produce interesting predictions and comparisons in models containing fixed spline effects. Consider the data in the following DATA step. Each of the 100 observations for the continuous response variable y is associated with one of two groups.

data spline;
  input group y @@;
  x = _n_;
  datalines;
 1    -.020 1    0.199 2    -1.36 1    -.026
 2    -.397 1    0.065 2    -.861 1    0.251
 1    0.253 2    -.460 2    0.195 2    -.108
 1    0.379 1    0.971 1    0.712 2    0.811
 2    0.574 2    0.755 1    0.316 2    0.961
 2    1.088 2    0.607 2    0.959 1    0.653
 1    0.629 2    1.237 2    0.734 2    0.299
 2    1.002 2    1.201 1    1.520 1    1.105
 1    1.329 1    1.580 2    1.098 1    1.613
 2    1.052 2    1.108 2    1.257 2    2.005
 2    1.726 2    1.179 2    1.338 1    1.707
 2    2.105 2    1.828 2    1.368 1    2.252
 1    1.984 2    1.867 1    2.771 1    2.052
 2    1.522 2    2.200 1    2.562 1    2.517
 1    2.769 1    2.534 2    1.969 1    2.460
 1    2.873 1    2.678 1    3.135 2    1.705
 1    2.893 1    3.023 1    3.050 2    2.273
 2    2.549 1    2.836 2    2.375 2    1.841
 1    3.727 1    3.806 1    3.269 1    3.533
 1    2.948 2    1.954 2    2.326 2    2.017
 1    3.744 2    2.431 2    2.040 1    3.995
 2    1.996 2    2.028 2    2.321 2    2.479
 2    2.337 1    4.516 2    2.326 2    2.144
 2    2.474 2    2.221 1    4.867 2    2.453
 1    5.253 2    3.024 2    2.403 1    5.498
;

The following statements produce a scatter plot of the response variable by group (Output 38.15.1):

proc sgplot data=spline;
   scatter y=y x=x / group=group name="data";
   keylegend "data"  / title="Group";
run;

Output 38.15.1 Scatter Plot of Observed Data by Group

The trends in the two groups exhibit curvature, but the type of curvature is not the same in the groups. Also, there appear to be ranges of values where the groups are similar and areas where the point scatters separate. To model the trends in the two groups separately and with flexibility, you might want to allow for some smooth trends in x that vary by group. Consider the following PROC GLIMMIX statements:

proc glimmix data=spline outdesign=x;
   class group;
   effect spl = spline(x);
   model y = group spl*group / s noint;
   output out=gmxout pred=p;
run;

The EFFECT statement defines a constructed effect named spl by expanding the x into a spline with seven columns. The group main effect creates separate intercepts for the groups, and the interaction of the group variable with the spline effect creates separate trends. The NOINT option suppresses the intercept. This is not necessary and is done here only for convenience of interpretation. The OUTPUT statement computes predicted values.

The "Parameter Estimates" table contains the estimates of the group-specific "intercepts," the spline coefficients varied by group, and the residual variance ("Scale," Output 38.15.2).

Because the B-spline coefficients for an observation sum to 1 and the model contains group-specific constants, the last spline coefficient in each group is zero. In other words, you can achieve exactly the same fit with the MODEL statement

model y = spl*group / noint;

or

model y = spl*group;

The following statements graph the observed and fitted values in the two groups (Output 38.15.3):

proc sgplot data=gmxout;
   series  y=p x=x / group=group name="fit";
   scatter y=y x=x / group=group;
   keylegend "fit" / title="Group";
run;

Output 38.15.3 Observed and Predicted Values by Group

Suppose that you are interested in estimating the mean response at particular values of x and in performing comparisons of predicted values. The following program uses ESTIMATE statements with nonpositional syntax to accomplish this:

proc glimmix data=spline;
   class group;
   effect spl = spline(x);
   model y = group spl*group / s noint;
   estimate 'Group 1, x=20' group 1    group*spl [1,1 20] / e;
   estimate 'Group 2, x=20' group 0  1 group*spl [1,2 20];
   estimate 'Diff at x=20 ' group 1 -1 group*spl [1,1 20] [-1,2 20];
run;

The first ESTIMATE statement predicts the mean response at in group 1. The E option requests the coefficient vector for this linear combination of the parameter estimates. The coefficient for the group effect is entered with positional (standard) syntax. The coefficients for the group*spl effect are formed based on nonpositional syntax. Because this effect comprises the interaction of a standard effect (group) with a constructed effect, the values and levels for the standard effect must precede those for the constructed effect. A similar statement produces the predicted mean at in group 2.

The GLIMMIX procedure interprets the syntax

group*spl [1,2 20] 

as follows: construct the spline basis at as appropriate for group 2; then multiply the resulting coefficients for these columns of the matrix with 1.

The final ESTIMATE statement represents the difference between the predicted values; it is a group comparison at .

The "Coefficients" table shows how the value supplied in the ESTIMATE statement was expanded into the appropriate spline basis (Output 38.15.4). There is no significant difference between the group means at (, Output 38.15.5).

The group comparisons you can achieve in this way are comparable to slices of interaction effects with classification effects. There are, however, no preset number of levels at which to perform the comparisons because x is continuous. If you add further values for the comparisons, a multiplicity correction is in order to control the familywise Type I error. The following statements compare the groups at values and compute simulation-based step-down-adjusted p-values. The results appear in Output 38.15.6. (The numeric results for simulation-based p-value adjustments depend slightly on the value of the random number seed.)

ods select Estimates;
proc glimmix data=spline;
   class group;
   effect spl = spline(x);
   model y = group spl*group / s;
   estimate 'Diff at x= 0' group 1 -1 group*spl [1,1  0] [-1,2  0],
            'Diff at x= 5' group 1 -1 group*spl [1,1  5] [-1,2  5],
            'Diff at x=10' group 1 -1 group*spl [1,1 10] [-1,2 10],
            'Diff at x=15' group 1 -1 group*spl [1,1 15] [-1,2 15],
            'Diff at x=20' group 1 -1 group*spl [1,1 20] [-1,2 20],
            'Diff at x=25' group 1 -1 group*spl [1,1 25] [-1,2 25],
            'Diff at x=30' group 1 -1 group*spl [1,1 30] [-1,2 30],
            'Diff at x=35' group 1 -1 group*spl [1,1 35] [-1,2 35],   
            'Diff at x=40' group 1 -1 group*spl [1,1 40] [-1,2 40],
            'Diff at x=45' group 1 -1 group*spl [1,1 45] [-1,2 45],
            'Diff at x=50' group 1 -1 group*spl [1,1 50] [-1,2 50],
            'Diff at x=55' group 1 -1 group*spl [1,1 55] [-1,2 55],            
            'Diff at x=60' group 1 -1 group*spl [1,1 60] [-1,2 60],
            'Diff at x=65' group 1 -1 group*spl [1,1 65] [-1,2 65],     
            'Diff at x=70' group 1 -1 group*spl [1,1 70] [-1,2 70],
            'Diff at x=75' group 1 -1 group*spl [1,1 75] [-1,2 75],
            'Diff at x=80' group 1 -1 group*spl [1,1 80] [-1,2 80] / 
            adjust=sim(seed=1) stepdown;
run;

There are significant differences at the low end and high end of the range. Notice that without the multiplicity adjustment you would have concluded at the 0.05 level that the groups are significantly different at . At the 0.05 level, the groups separate significantly for and .