The Garbage Class Mixed Logit Model: Accounting for Low-Quality Response Patterns in Discrete Choice Experiments.

Area of Science:

• Behavioral Economics
• Econometrics
• Marketing Science

Background:

• Discrete choice experiments (DCEs) are widely used to model consumer preferences.
• Identifying and handling respondents with low data quality is crucial for robust DCE analysis.
• Traditional methods for screening low-quality data can be labor-intensive and ambiguous.

Purpose of the Study:

• To introduce and evaluate the garbage class mixed logit (MIXL) model as a streamlined approach for managing low-quality data in DCEs.
• To demonstrate the utility of garbage classes within the MIXL framework.

Main Methods:

• The study reanalyzed four existing DCE datasets originally analyzed with standard MIXL models.
• A garbage class was incorporated into the MIXL model to identify respondents exhibiting random choice behavior.
• The performance of the garbage class MIXL model was compared to manual screening methods based on root likelihood tests.

Main Results:

• Garbage class MIXL models effectively replicate the results of manually excluding low-quality respondents.
• This approach reduces the effort and ambiguity associated with traditional data quality screening.
• The analysis demonstrated that garbage class MIXL models can identify respondents with random choice patterns.

Conclusions:

• Incorporating a garbage class into MIXL models automatically removes the influence of random responders.
• This method provides an estimate of low-quality respondents without manual intervention.
• The garbage class MIXL model presents a practical and efficient alternative for enhancing DCE data quality analysis.