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

[Epidemiologic and statistical interaction models and consequences for regression analysis].

W J Stronegger1, A Berghold, G U Seeber

  • 1Institut für Sozialmedizin, Universität Graz. willibald.stronegger@kfunigraz.ac.at

Sozial- Und Praventivmedizin
|February 20, 1999
PubMed
Summary
This summary is machine-generated.

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

On the correlations of biomechanical properties of super-imposed temporal tissue layers and their age-, sex-, side- and post-mortem interval dependence.

Journal of biomechanics·2021
Same author

Ventilation time and prognosis after stroke thrombectomy: the shorter, the better!

European journal of neurology·2020
Same author

Serum sclerostin levels in renal cell carcinoma patients with bone metastases.

Scientific reports·2016
Same author

Anaphylaxis with clonal mast cells in normal looking skin - a new entity?

Allergy·2015
Same author

LGR5 rs17109924 is a predictive genetic biomarker for time to recurrence in patients with colon cancer treated with 5-fluorouracil-based adjuvant chemotherapy.

The pharmacogenomics journal·2015
Same author

Evaluation of prognostic scoring systems for spinal metastases in 196 patients treated during 2005-2010.

European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society·2014
Same journal

Time trends in cancer mortality in central Serbia.

Sozial- und Praventivmedizin·2007
Same journal

Findings from the Global Youth Tobacco Survey (GYTS) in Czech Republic, Hungary, Poland and Slovakia--smoking initiation, prevalence of tobacco use and cessation.

Sozial- und Praventivmedizin·2007
Same journal

Reference values for serum lipids and lipoproteins in Spanish adolescents: the AVENA study.

Sozial- und Praventivmedizin·2007
Same journal

Tobacco use among youth and related characteristics, Turkey.

Sozial- und Praventivmedizin·2007
Same journal

Adolescents' education and their diet recorded by 7-day food records.

Sozial- und Praventivmedizin·2007
Same journal

Towards the end of the Spanish paradox?

Sozial- und Praventivmedizin·2007
See all related articles

Understanding interaction in epidemiological studies is crucial. This study highlights how assuming multiplicative models can bias results, especially with additive effects, and suggests using generalized linear models for accurate analysis.

Area of Science:

  • Epidemiology
  • Biostatistics
  • Causal Inference

Context:

  • Standard epidemiological analyses often assume multiplicative interaction models (e.g., Poisson, logistic regression).
  • This assumption can lead to confusion and bias, particularly when the underlying data structure is additive.
  • Causal models of disease etiology suggest non-multiplicative concepts of no interaction are plausible.

Purpose:

  • To illustrate the asymptotic bias ('interaction-bias') in main effect estimates when using inappropriate multiplicative models.
  • To demonstrate how epidemiological study design and causal models influence interaction structures.
  • To advocate for the use of generalized linear models for analyzing non-multiplicative interaction structures.

Summary:

  • The study highlights the 'interaction-bias' arising from the default multiplicative parameterization in event data analysis.

Related Experiment Videos

  • It shows that additive or other non-multiplicative interaction concepts are supported by empirical and causal evidence.
  • Appropriate model selection requires considering both study design and underlying causal models for accurate interaction analysis.

    • Impact:

    • Provides a clearer understanding of interaction definitions (risk, rate, odds) in epidemiological research.
    • Offers a methodological framework using generalized linear models to correctly analyze non-multiplicative interactions.
    • Aims to reduce analytical errors and improve the validity of findings in epidemiological studies concerning exposure interactions.