Figure 6.15 shows the Parameter Estimates report obtained by running the script MaxDiff for Flavor in Potato Chip Responses.jmp.
Figure 6.15 Parameter Estimates Report
An estimate of the parameter associated with the corresponding term. In discrete choice experiments, parameter estimates are sometimes referred to as part-worths. Each part-worth is the coefficient of utility associated with the given term. By default, these estimates are based on the Firth bias-corrected maximum likelihood estimators and therefore are considered to be more accurate than MLEs without bias correction.
The following fit statistics are shown as part of the report and can be used to compare models: AICc (corrected Akaike’s Information Criterion), BIC (Bayesian Information Criterion), − 2*LogLikelihood, and − 2*Firth LogLikelihood. See Statistical Details in the Fitting Linear Models book for details on the first three of these measures.
The − 2*Firth LogLikelihood value is included in the report only when the Firth Bias-adjusted Estimates check box is checked in the launch window. This option is checked by default.