|
|
 |
| Constraining Parameters with "Constra... |
 
|
| Message/Author |
|
|
|
|
I am trying to constrain regression parameters to match uneven spacing of observations across participants. The goal is to have exponential decay for periods greater than one year.
I created constraint variables:
CONSTRAINT =
one_two
two_three
three_four
four_five
five_six
t5
t6
t7
t8
t9;
DEFINE:
t5 = exp((-1*one_two)+1);
t6 = exp((-1*two_three)+1);
t7 = exp((-1*three_four)+1);
t8 = exp((-1*four_five)+1);
t9 = exp((-1*five_six)+1);
For my outcomes at each of the 5 timepoints, I named the parameter, then used the constraints to set them equal:
!T5
[bmi_5*];
bmi_5*;
bmi_5 ON bmi_3 fev1_3 (sb5 ib5);
[...]
MODEL CONSTRAINT:
!BMI stability
new(sb);
sb = sb5*t5;
sb = sb6*t6;
sb = sb7*t7;
sb = sb8*t8;
sb = sb9*t9;
This results in a Fatal Error.
Is there a way that I can incorporate these weights into my regression parameters and set them equal? |
|
|
|
|
Try this
MODEL CONSTRAINT:
new(sb);
sb5 = sb/t5;
sb6 = sb/t6;
sb7 = sb/t7;
sb8 = sb/t8;
sb9 = sb/t9; |
|
|
|
|
| Thank you! That works. |
|
|
| Back to top |
|
|
