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

Effects of randomization methods on statistical inference in disease cluster detection.

Colleen C McLaughlin1, Francis P Boscoe

  • 1New York State Cancer Registry, New York State Department of Health, Corning Tower Room 536, Empire State Plaza, Albany, NY 12237, USA.

Health & Place
|January 13, 2006
PubMed
Summary

Randomizing disease rates in cluster analysis can bias results, especially with varying variances. Randomizing case counts proportional to population offers unbiased statistical inference for disease cluster detection.

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

Racial and Ethnic Disparities in Care for Pediatric Sleep-Disordered Breathing.

OTO open·2026
Same author

Ethnic Enclave Residence and Breast, Cervical, and Colorectal Cancer Incidence among Asian American Adults, 2006 to 2017.

Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology·2026
Same author

Combined Effects of Race and Ethnicity With Rurality on Tonsillectomy Rates Among Children Enrolled in Medicaid.

Academic pediatrics·2026
Same author

Inferring Unknown Race in Central Cancer Registries.

Journal of registry management·2025
Same author

Health Care Utilization Prior to Ovarian Cancer Diagnosis in Publicly Insured Individuals in New York State.

Journal of registry management·2024
Same author

Asian American Enclaves and Healthcare Accessibility: An Ecologic Study Across Five States.

American journal of preventive medicine·2023

Area of Science:

  • Epidemiology
  • Biostatistics
  • Geographic Information Systems (GIS)

Background:

  • Monte Carlo methods are standard for assessing disease cluster significance via outcome permutation.
  • Heterogeneous variance in disease rates across geographic units introduces bias when randomizing rates.
  • This bias leads to under-ascertainment of clusters in urban and over-ascertainment in rural areas.

Purpose of the Study:

  • To evaluate the bias introduced by randomizing disease rates in cluster analysis.
  • To propose and compare an alternative method: randomizing case counts proportional to population.
  • To assess the impact of these methods on statistical inference for disease clusters.

Main Methods:

  • Comparison of two Monte Carlo permutation approaches: randomizing disease rates versus randomizing case counts.

Related Experiment Videos

  • Utilized the local Moran's I statistic for cluster detection.
  • Applied methods to county-level prostate cancer mortality data (US) and ZIP-Code level incidence data (New York State).

    • Main Results:

    • Randomizing disease rates leads to biased p-values, overestimating in low-variance areas and underestimating in high-variance areas.
    • Randomizing case counts proportional to population preserves the variance structure, yielding unbiased statistical inference.
    • The study identified differential cluster ascertainment between urban and rural areas using the rate randomization method.

    Conclusions:

    • Randomizing disease rates is inappropriate for heterogeneous geographic variance, causing biased disease cluster detection.
    • Randomizing case counts proportional to population is a more robust method for unbiased cluster significance assessment.
    • This finding has implications for accurate identification of disease hotspots and resource allocation.