A Guide to Observable Differences in Stated Preference Evidence.

Area of Science:

• Health Preference Research
• Behavioral Economics
• Public Health Policy

Background:

• Health preference studies often hypothesize parameter or choice prediction differences based on observable factors.
• Discrete choice experiments are key tools for testing and estimating these observable differences.
• This guide provides a framework for exploring and corroborating such differences in health preference evidence.

Purpose of the Study:

• To guide researchers in health preference research on identifying and corroborating observable differences.
• To demonstrate a three-step analytical process: exploratory data analysis, confirmatory data analysis, and evidence interpretation.
• To apply this approach to 2020 US COVID-19 vaccination preferences using dual samples.

Main Methods:

• The study employed a three-step analytical process: exploratory data analysis, confirmatory data analysis, and interpretation/dissemination.
• A dual-sample approach was used, with separate sources for exploratory and confirmatory data.
• The methodology is applicable to both dual and split-sample designs.

Main Results:

• Confirmatory analysis did not reject 10 out of 17 null hypotheses from the exploratory analysis (p < 0.05).
• Observable differences were identified across demographic, socioeconomic, and geographic factors.
• Political independents and individuals never vaccinated against influenza were least likely to be vaccinated for COVID-19 (0.838 and 0.872 probability, respectively).

Conclusions:

• Mastering the identification and corroboration of observable differences is crucial for health preference researchers.
• This skill enables more complex analyses, such as those involving latent classes or random parameters.
• The guide concludes with reflective questions for researchers conducting or reviewing analyses of observable differences.