Here we are working with some artificial multinomial data
    generated in the EVIEWS 4.0 manual.  See Chapter 18 on
    "The Log Likelihood (LogL) Object" and in particular the
    discussion of the EVIEWS program mlogit1.prg on pp. 470 - 
    473.  This program can be obtained from the EVIEWS example
    programs file located at
         c:/Program Files/EVIEWS4/Example Files/mlogit1.prg       

    To make the problem interesting I will give the artificial data
    a "story."  Let us assume that we have observations on two
    characteristics (X1 and X2, say age and income if you like)
    for 1000 consumers in a given location who choose to shop for
    groceries at either store 1 (st = 1), or store 2 (st = 2),
    or store 3 (st = 3).  Here we are going to use the SAS procedure
    Logistic to fit an unordered multinomial logit model to the data
    where st is the unranked multinomial variable and x1 and x2 are
    the explanatory variables.  

    Note the data dd2, dd3, x1 and x2 come directly from the
    the mlogit1.wf1 data set.  The mlogit1.prg EVIEWS 4.0
    program does the same analysis of the data.