[Limdep Nlogit List] How to simulate impact of removal of alternative?

[Mindaugas Zickus]

Hello
 
I wonder if anyone could suggest how to simulate the impact of removal of one of the choice alternatives using mixed logit model estimated on panel data. Data is an unbalanced panel of households chosing one of the 4 brands during the store visits. The mixed logit model specification should allow estimating the unproportional impact on each of the remaining alternatives (i.e. flexible substitution).
 
I'm trying to estimate random intercept (heterogeneous taste) mixed model with code:
 Nlogit 
; Lhs = Purchase 
; Choices=Brand1,Brand2,Brand3,Brand4
; Rhs = Int1, Int2, Int3
; RPL 
; Cluster=Pannelis
; Correlated
; Pts=25
; Fcn= Int1(n),Int2(n),Int3(n) $
 
 
Thanks, Mindis
 
 
first 30 data lines:
 
PANNELIS WEEK PURCHASE BRAND INT1 INT2 INT3
9 11 0 1 0 0 1
9 11 0 2 0 1 0
9 11 1 3 1 0 0
9 11 0 4 0 0 0
9 29 0 1 0 0 1
9 29 1 2 0 1 0
9 29 0 3 1 0 0
9 29 0 4 0 0 0
9 31 0 1 0 0 1
9 31 1 2 0 1 0
9 31 0 3 1 0 0
9 31 0 4 0 0 0
9 43 0 1 0 0 1
9 43 1 2 0 1 0
9 43 0 3 1 0 0
9 43 0 4 0 0 0
9 45 0 1 0 0 1
9 45 1 2 0 1 0
9 45 0 3 1 0 0
9 45 0 4 0 0 0
9 49 0 1 0 0 1
9 49 1 2 0 1 0
9 49 0 3 1 0 0
9 49 0 4 0 0 0
9 51 0 1 0 0 1
9 51 1 2 0 1 0
9 51 0 3 1 0 0
9 51 0 4 0 0 0
11 3 0 1 0 0 1
11 3 0 2 0 1 0
11 3 1 3 1 0 0
11 3 0 4 0 0 0
11 11 0 1 0 0 1
11 11 0 2 0 1 0
11 11 1 3 1 0 0
11 11 0 4 0 0 0
.
.
.
and so on