COMMENT	spss code for Example 1 .
COMMENT	partit is the partition prime condition.
COMMENT	elicit is whether the judged probabilities were imprecise or precise.
MODEL PROGRAM B01 = -1.7 G01 = -3.4 Q = -1.2 Q1 = -0.55 Q2 = 0.47  .
COMPUTE #DV = pyes .
COMMENT	Composition submodel .
COMPUTE #C = 1/(1+EXP(Q + Q1*partit + Q2*elicit)) .
COMMENT	Location submodel . 
COMPUTE #M1 = EXP(B01)/(1+EXP(B01)) .
COMPUTE #M2 = 1/2 .
COMPUTE #MU = #C*#M1 + (1-#C)*#M2  .
COMMENT	Dispersion submodel .
COMPUTE PHI1 = EXP(-G01) .
COMPUTE PHI2 = EXP(8) .
COMPUTE PRED_ = #MU .
COMPUTE RESID_ = #DV  - PRED_ .
COMPUTE LL = LN(MAX(0.001, #C*PDF.BETA(#DV, #M1*PHI1, MAX(0.001, PHI1 - #M1*PHI1))+(1-#C)* PDF.BETA(#DV, #M2*PHI2, MAX(0.001, PHI2 -
    #M2*PHI2)))) .
COMPUTE LOSS_  = -LL .
COMMENT	To get bootstrap estimates, put /BOOTSTRAP n statement before /CRITERIA statement.
COMMENT	To save predicted values, residuals and gradient put /SAVE PRED RESID DERIVATIVES before /CRITERIA statement.
CNLR pyes
  /OUTFILE='C:\SPSSFNLR.TMP'
  /PRED PRED_
  /LOSS LOSS_
  /CRITERIA STEPLIMIT 2 ISTEP 1E+20 .