Strategies of model construction for the analysis of judgment data

This paper is concerned with the types of models researchers use to
analyze empirical data in the domain of social judgments and decisions. Models
for the analysis of judgment data may be divided into two classes depending on
the criteria they optimize: Optimizing an internal (mathematical) criterion function
with the aim to minimize the discrepancy of values predicted by the model from
obtained data or incorporating a substantive underlying theory into the model where
model parameters are not only formally defined, but represent specified components
of judgments. Results from applying models from both classes to empirical data
exhibit considerable differences between the models in construct validity, but not in
empirical validity. It may be concluded that any model for the analysis of judgment
data implies the selection of a formal theory about judgments. Hence, optimizing a
mathematical criterion function does not induce a non-theoretical rationale or neutral
tool. As a consequence, models satisfying construct validity seem superior in
the domain of judgments and decisions.