st: question on nested logit

["Data Analytics Corp." <walt@dataanalyticscorp.com>]

Hi,

I'm experimenting with the nested logit setup for a coffee demand study 
I want to do.  Here's a sample of the data:

+-----------------------------------------------------------------------------------------------------+
     | id   choice      form       flavor   size   price brandc~e      
brand      FLAVORX     FORMX    SIZEX |
|-----------------------------------------------------------------------------------------------------|
  1. |  1        0         BagGr    UnLgt     16    2.25 4 Folgers   
UNFLAVORED    BAGGED   MEDIUM |
  2. |  1        0         BagWh    UnMed     20       3 7 SBUX   
UNFLAVORED    BAGGED    LARGE |
  3. |  1        0         Can         UnDK     18 2.75          6 
Classico   UNFLAVORED    CANNED   MEDIUM |
  4. |  1        0         Pods        Hazel     10     3.1 1         
DD     FLAVORED      PODS    SMALL |
  5. |  1        0       InstCan      Van     24    1.75 8 Store     
FLAVORED   INSTANT    LARGE |
|-----------------------------------------------------------------------------------------------------|
  6. |  1        0    InstSt    Other      4     2.5 5 Peet     
FLAVORED   INSTANT    SMALL |
  7. |  1        0       Mix    UnLgt     12    3.25          3 
Maxwell   UNFLAVORED     MIXES   MEDIUM |
  8. |  1        1       Liq    UnMed      8     1.5 2 Godiva   
UNFLAVORED    LIQUID    SMALL |
  9. |  2        0     BagGr     UnDK     16    2.25          4 
Folgers   UNFLAVORED    BAGGED   MEDIUM |
 10. |  2        0     BagWh    Hazel     20       3 7 SBUX     
FLAVORED    BAGGED    LARGE |
+-----------------------------------------------------------------------------------------------------+

The data are read from a text file using a Do File which also creates 
the nested logit tree:

#delimit ;
insheet using "C:\Temp\coffee.txt", clear;
label drop _all;
nlogitgen FLAVORX = flavor(UNFLAVORED: UnLgt | UnMed | UnDK, FLAVORED: 
Hazel | Van | Other);
nlogitgen FORMX = form(BAGGED: BagGr | BagWh, CANNED: Can, INSTANT: 
InstCan | InstSt, MIXES: Mix, LIQUID: Liq, PODS: Pods);
nlogitgen SIZEX = size(SMALL: 4 | 8 | 10, MEDIUM: 12 | 16 | 18, LARGE: 
20 | 24);
nlogittree brand FORMX FLAVORX SIZEX, choice(choice);

Stata returns an error message:

. nlogittree brand FORMX FLAVORX SIZEX, choice(choice);
invalid tree structure; input bottom level first
r(198);

Why?  No matter how I arrange FORMX, FLAVORX, and SIZEX, I get an 
error.  I want brand to be at the bottom of the tree and it's specified 
first, as it should be.  Everything seems correct.  But obviously I'm 
missing something.  Any help?

I'm using Stata 11 on a Windows PC.

Thanks,

Walt
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/