|
|
| Message/Author |
|
|
|
|
Hi,
about example 5.23 :
1. is there a technical description of how the estimation of a parameter moderated by a continuous variable works? (for instance, let's assume that pihat were continuous rather than={0, .25, .5, .75, 1})
2. the output provides the estimated value a of pihat on the cov parameter. Any way of obtaining the value of cov (Y1 WITH Y2) and its statistical test?
thanks, |
|
|
|
|
1. We have nothing written (although perhaps Mike Neale has done it); it's a pretty straightforward parameter restriction. A continuous variable is not problematic. For instance, in our RMA book we use a continuous x variable to model residual variance heteroscedasticity as
resvar = exp(a + b*x);
So that is imposing the structure of a log linear function of x.
2. Try
New(covest);
covest = cov;
or replace cov with the RHS expression for cov. |
|
|
| Back to top |
|
|
