Best-subset instrumental variable selection method using mixed integer optimization with applications to health-related quality of life and education–wage analyses
2024

Best-subset Instrumental Variable Selection Method for Health and Education Analysis

Sample size: 1816 publication 10 minutes Evidence: high

Author Information

Author(s): Qasim Muhammad, Månsson Kristofer, Balakrishnan Narayanaswamy

Primary Institution: Jönköping International Business School, Jönköping University, Sweden

Hypothesis

Can the best-subset instrumental variable (BSIV) method using mixed integer optimization improve causal effect estimation in the presence of invalid instruments?

Conclusion

The BSIV method outperforms traditional estimation methods in estimating causal effects when instruments may be invalid.

Supporting Evidence

  • The BSIV method was shown to yield substantial improvements over existing estimators in terms of bias and median squared error.
  • Monte Carlo simulations demonstrated the robustness of the BSIV method across various scenarios.
  • Real-life applications confirmed the effectiveness of the BSIV method in estimating causal effects.

Takeaway

This study shows a new way to pick the best tools to understand how things like education and health affect each other, even when some tools might not work well.

Methodology

The study uses mixed integer optimization to select valid and invalid instrumental variables for linear regression models, validated through Monte Carlo simulations and real-life datasets.

Potential Biases

Potential bias exists if the instruments are not properly validated or if the model assumptions are not met.

Limitations

The method may not perform well if a large portion of instruments are invalid or if the assumptions of the model are violated.

Participant Demographics

The analysis included data from American high school graduates tracked since 1957, totaling 1816 observations.

Statistical Information

P-Value

p<0.001

Confidence Interval

95% CI (−0.0071, −0.0051)

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.21203/rs.3.rs-5550004

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication