Combining List Experiment and Direct Question Estimates of Sensitive Behavior Prevalence

Area of Science:

• Social Sciences
• Survey Methodology
• Statistics

Background:

• Direct questioning in surveys can lead to untruthful answers for sensitive topics.
• List experiments (item count technique) reduce bias but increase sampling variability.
• A need exists for methods balancing bias reduction and statistical efficiency.

Purpose of the Study:

• To develop a novel nonparametric estimator combining list experiments and direct questioning.
• To enhance the efficiency and reliability of estimating sensitive behavior prevalence.
• To introduce new placebo tests for validating list experiment assumptions.

Main Methods:

• Developed a combined nonparametric estimator integrating list experiment and direct questioning data.
• Proved asymptotic efficiency of the new estimator compared to standard methods.
• Derived nonparametric placebo tests using direct questioning information for assumption validation.

Main Results:

• The combined estimator is asymptotically more efficient than the difference-in-means estimator.
• Wald-type confidence intervals provide a basis for statistical inference.
• Placebo tests effectively assess identifying assumptions of list experiments.
• An original survey experiment demonstrated the estimator and tests' effectiveness.

Conclusions:

• Combining list experiments and direct questioning offers a superior method for sensitive survey data.
• The new estimator and placebo tests improve the accuracy and validity of prevalence estimates.
• This integrated approach provides researchers with more robust tools for social science research.