ABSTRACT

We perform judgment tasks in our daily life. A consumer makes a judgment about which car to buy after considering various cues (or information) about the car like model, year, mileage, price, re-sale value, top speed, etc. A human resource professional makes a judgment about the starting salary of an employee by analyzing different cues related to the starting position, employment location, company budget, bonus options, etc.

A lot of work has already been done to understand judgment tasks in various laboratory and field studies using regression models, known as lens models. However, there is no consensus among researchers regarding how many cues should be provided to the human judge to make a judgment. Hence, to understand the effect of task characteristics on judgment, we tested the effect of change in the cardinality of the cue set on human judgment.

In addition to this, the relationship between representation of the cues and the judgment options with the human judgment is not well understood. We believe there is a need to understand how these variations in the task characteristics; representation of cues and judgment options, affect human judgment.

Hence this work tried to answer the following research questions by conducting human subject experiments in a controlled laboratory setting:

1.     whether changing the cardinality of the cue set affects human judgment?

2.     whether changing the representation of the cues affect human judgment? and

3.     whether changing the representation of judgment options affect human judgment?

In addition, researchers have done studies in which judgments have been modeled linearly, where cues act as predictors and judgments act as response variables in a regression using a lens model. However, there is no empirical evidence whether lens model parameters can be compared when task characteristics are changed. In the past, many meta-analytical studies compare different lens model studies without considering the sensitivity of lens model parameters to cue cardinality, cue representation, and representation of the judgment options.

Therefore, in addition to human experiments, a Monte Carlo simulation was developed to check whether lens model is sensitive to the different task characteristics or not?

The findings from the simulation suggest that lens model is insensitive to the task characteristics and thus can be used to understand different judgment models. From human experiment studies, we found evidence that change in cardinality of the cue set, representation of the cues and judgment options do not affect human judgment. These results are a first step towards understanding human judgment and judgment tasks with varying task characteristics using lens model.