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

Imputation of missing longitudinal data: a comparison of methods.

Jean Mundahl Engels1, Paula Diehr

  • 1Departments of Biostatistics and Health Services, University of Washington, 1959 Northeast Pacific Avenue, Box 357232, Seattle, WA 98195, USA. mundahl@u.washington.edu

Journal of Clinical Epidemiology
|October 22, 2003
PubMed
Summary

Imputing missing data in longitudinal studies is crucial for accurate results. Methods using a person's own past and future data performed best for estimating missing health information in older adults.

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

Predicting Future Years of Life, Health, and Functional Ability: A Healthy Life Calculator for Older Adults.

Gerontology & geriatric medicine·2017
Same author

Disability and recovery of independent function in obstructive lung disease: the cardiovascular health study.

Respiration; international review of thoracic diseases·2014
Same author

Effects of respiratory and non-respiratory factors on disability among older adults with airway obstruction: the Cardiovascular Health Study.

COPD·2013
Same author

Transitions among Health States Using 12 Measures of Successful Aging in Men and Women: Results from the Cardiovascular Health Study.

Journal of aging research·2012
Same author

Persistence and remission of musculoskeletal pain in community-dwelling older adults: results from the cardiovascular health study.

Journal of the American Geriatrics Society·2012
Same author

Improving low-wage, midsized employers' health promotion practices: a randomized controlled trial.

American journal of preventive medicine·2012

Area of Science:

  • Gerontology
  • Biostatistics
  • Longitudinal Studies

Background:

  • Missing data is common in longitudinal studies, potentially biasing results and reducing statistical power.
  • Data imputation is a strategy to address missing information and create more complete datasets.
  • Older adults are a key population where longitudinal health data is vital.

Purpose of the Study:

  • To evaluate the effectiveness of 14 different data imputation methods.
  • To compare imputation performance for key health indicators: depression, weight, cognitive function, and self-rated health.
  • To identify optimal imputation strategies for longitudinal data in older adults.

Main Methods:

  • A novel approach treated known future values as "missing" for imputation comparison.

Related Experiment Videos

  • Evaluated methods based on root mean square error, mean absolute deviation, bias, and relative variance.
  • Compared imputation techniques using longitudinal data from a cohort of older adults.

    • Main Results:

    • Most imputation methods showed bias, estimating "missing" values as healthier than reality.
    • Estimates derived from imputation methods often had lower variance than observed data.
    • Imputation methods utilizing a person's pre- and post-missing data outperformed others, followed by methods using only pre-missing data.

    Conclusions:

    • Imputation methods relying on general population data (e.g., sample mean) performed poorly.
    • For longitudinal studies with a declining health trend, imputing missing data using individual longitudinal data is recommended.
    • Prioritizing person-specific longitudinal data improves the accuracy of missing data imputation in older adult cohorts.