Beyond logistic regression: structural equations modelling for binary variables and its application to investigating unobserved confounders
2006

Using Structural Equation Modeling for Binary Data

Sample size: 5000 publication 10 minutes Evidence: moderate

Author Information

Author(s): Emil Kupek

Primary Institution: National Perinatal Epidemiology Unit, Institute of Health Sciences, University of Oxford, UK

Hypothesis

Can Yule's transformation of odds ratios improve the analysis of binary data using structural equation modeling?

Conclusion

Yule's transformation of odds ratios can significantly enhance the analysis of binary data through structural equation modeling.

Supporting Evidence

  • The study demonstrated that SEM can recover the model structure effectively.
  • SEM suggested significant influences of latent confounding variables not detectable by logistic regression.
  • The classification performance of SEM was broadly similar to that of logistic regression.

Takeaway

This study shows that a special math method can help us understand how different things are related when we only have yes or no answers.

Methodology

The study used simulated data and real obstetric data to compare the performance of structural equation modeling with logistic regression.

Potential Biases

Potential biases may arise from the assumptions made in the modeling process, particularly regarding the treatment of categorical variables.

Limitations

The study's findings may not be generalizable beyond the specific models tested, and the approximation of Yule's Q to Pearson's correlation may not hold in all cases.

Participant Demographics

The real data set included 10,574 multiparous women with singleton pregnancies.

Statistical Information

P-Value

p<0.001

Confidence Interval

(1.14, 1.29)

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.1186/1471-2288-6-13

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication