Investigating attribute non-attendance and its consequences in choice experiments with latent class models

Area of Science:

• Behavioral Economics
• Health Economics
• Econometrics

Background:

• Discrete Choice Experiments (DCEs) are widely used to elicit preferences.
• A growing concern is that respondents may not adhere to preference axioms, employing strategies like attribute non-attendance (ANA).
• ANA involves respondents ignoring one or more attributes during choice tasks.

Purpose of the Study:

• To investigate attribute non-attendance (ANA) in a DCE among healthcare providers in Ghana.
• To evaluate healthcare providers' resistance to changes in clinical guidelines.
• To demonstrate the application of latent class models in accounting for ANA strategies.

Main Methods:

• A Discrete Choice Experiment (DCE) was administered to healthcare providers.
• Latent class models were employed in a step-wise approach to identify and model ANA strategies.
• The study analyzed the impact of ANA on model fit and parameter estimates.

Main Results:

• Less than 3% of respondents considered all attributes; most focused on one or two.
• Accounting for ANA significantly improved the model's goodness-of-fit.
• ANA affected the magnitude of some coefficient and willingness-to-pay estimates.
• Predicted probabilities remained consistent between models with and without ANA.

Conclusions:

• Attribute non-attendance (ANA) is prevalent in DCEs among healthcare providers.
• While ANA impacts model fit and estimates, it may not bias overall policy predictions.
• Further research is needed to confirm the robustness of these findings across different contexts.