Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A method to automate probabilistic sensitivity analyses of misclassified binary variables.

Matthew P Fox1, Timothy L Lash, Sander Greenland

  • 1Department of International Health, Boston University School of Public Health, Boston, MA, USA. mfox@bu.edu

International Journal of Epidemiology
|September 21, 2005
PubMed
Summary

Misclassification bias in studies is common, but its impact is rarely quantified. This new method uses probabilistic sensitivity analysis to estimate the likely effects of misclassification, providing more accurate study results.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

The complexities of comprehensive exams in epidemiology.

American journal of epidemiology·2026
Same author

Paralysis in public health: evidence as an alibi? - Authors' reply.

The Lancet. Public health·2026
Same author

Quantitative bias analysis for time to event data needs good validation data.

American journal of epidemiology·2026
Same author

Combined Oral Contraceptives and Stroke Risk in Individuals With Migraine With Aura.

Obstetrics and gynecology·2026
Same author

Hepatitis B and C Prevention, Screening and Diagnostic Services at HIV Treatment Sites: International epidemiology Databases to Evaluate AIDS.

Journal of acquired immune deficiency syndromes (1999)·2026
Same author

One- and Two-Year Outcomes of Switching to a Dolutegravir-Based ART Regimen: An Observational Emulation of Sequential Target Trials.

Open forum infectious diseases·2026

Area of Science:

  • Epidemiology
  • Biostatistics
  • Health Research Methods

Background:

  • Misclassification bias is a pervasive issue in scientific research.
  • Quantifying the uncertainty associated with misclassification bias is infrequently performed.

Purpose of the Study:

  • To introduce a novel method for probabilistic sensitivity analysis.
  • To quantify the potential impact of misclassification bias on study outcomes.

Main Methods:

  • Developed a method to reconstruct data assuming correct classification based on sensitivity and specificity.
  • Created an accompanying SAS macro to implement the analysis.
  • The method generates simulation intervals accounting for systematic and random error.

Main Results:

Related Experiment Videos

  • Applied the method to a study on resin exposure and lung cancer.
  • Compared results with conventional analyses and original sensitivity analyses.
  • Demonstrated the ability to incorporate uncertainty from misclassification.

Conclusions:

  • Investigators can present study findings with quantified uncertainty due to misclassification.
  • This approach helps avoid the presentation of misleadingly precise results.
  • Enhances the transparency and reliability of research findings.