Related Experiment Video
Updated: May 20, 2026
06:55
Inverse Probability of Treatment Weighting (Propensity Score) using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
Comparison of methods for imputing ordinal data using multivariate normal imputation: a case study of non-linear
Katherine J Lee1, John C Galati, Julie A Simpson
1Clinical Epidemiology and Biostatistics Unit, Murdoch Childrens Research Institute, Melbourne, Australia. katherine.lee@mcri.edu.au
Statistics in Medicine
|July 25, 2012
Summary
Multiple imputation for missing data using multivariate normality (MVN) can bias ordinal variable associations. Imputing as continuous distorts non-linear relationships, while indicator methods preserve them but not marginal distributions.
Area of Science:
- Statistics
- Biostatistics
- Data Science
Background:
- Multiple imputation is a popular technique for handling missing data.
- Markov chain Monte Carlo (MCMC) assuming multivariate normality (MVN) is a common imputation approach.
- Imputing non-normal categorical variables using MVN is challenging, with various methods proposed but no clear preference.
Purpose of the Study:
- To explore and compare methods for imputing ordinal variables within an MVN imputation framework.
- To evaluate the impact of different imputation strategies on estimating non-linear exposure-outcome associations.
Main Methods:
- Investigated imputing ordinal variables as continuous and as a set of indicators.
- Examined various category assignment (rounding) methods for continuous imputation.
- Introduced a novel approach combining continuous imputation with mean indicator assignment.
- Compared methods using a real dataset with 50% missingness in an ordinal exposure, assuming data missing completely at random.
Main Results:
- Imputing ordinal variables as continuous biased the non-linear exposure-outcome association towards linearity, regardless of rounding method.
- Imputing ordinal variables using indicators preserved the non-linear association.
- Indicator imputation methods did not preserve the marginal distribution of the ordinal variable.
Conclusions:
- Treating ordinal variables as continuous in MVN imputation can significantly bias exposure-outcome association estimates when non-linear relationships exist.
- Further research is required to establish optimal imputation methods for ordinal and nominal variables using MVN.
Related Concept Videos
01:26Longitudinal Studies
Longitudinal studies are also widely used in other medical and social science fields. For instance, in cardiovascular research, they can monitor patients' health over decades to identify risk factors for heart disease, such as high cholesterol or smoking, and evaluate the long-term effectiveness of preventive measures. Similarly, in mental health studies, researchers might follow individuals from adolescence into adulthood to understand the development and progression of conditions like...
01:16Statistical Inference Techniques in Hypothesis Testing: Parametric Versus Nonparametric Data
Statistical inference techniques, paramount in hypothesis testing, differentiate into two broad categories: parametric and nonparametric statistics.
Parametric statistics, as the name suggests, assumes that data follow a specific distribution, often a normal distribution. This assumption enables robust hypothesis testing and estimation. Parametric methods, like the Student's t-test or Goodness-of-fit test, are frequently employed in biostatistics due to their robustness. For instance, comparing...
Parametric statistics, as the name suggests, assumes that data follow a specific distribution, often a normal distribution. This assumption enables robust hypothesis testing and estimation. Parametric methods, like the Student's t-test or Goodness-of-fit test, are frequently employed in biostatistics due to their robustness. For instance, comparing...
01:25Statistical Methods for Analyzing Epidemiological Data
Epidemiological data primarily involves information on specific populations' occurrence, distribution, and determinants of health and diseases. This data is crucial for understanding disease patterns and impacts, aiding public health decision-making and disease prevention strategies. The analysis of epidemiological data employs various statistical methods to interpret health-related data effectively. Here are some commonly used methods:
00:55Ordinal Level of Measurement
The way a set of data is measured is called its level of measurement. Correct statistical procedures depend on a researcher being familiar with levels of measurement. For analysis, data are classified into four levels of measurement—nominal, ordinal, interval, and ratio.
Data measured using an ordinal scale are similar to nominal scale data, but there is one major difference. The ordinal scale data can be ordered. An example of ordinal scale data is a list of the top five national parks in the...
Data measured using an ordinal scale are similar to nominal scale data, but there is one major difference. The ordinal scale data can be ordered. An example of ordinal scale data is a list of the top five national parks in the...
01:20Comparing the Survival Analysis of Two or More Groups
Survival analysis is a cornerstone of medical research, used to evaluate the time until an event of interest occurs, such as death, disease recurrence, or recovery. Unlike standard statistical methods, survival analysis is particularly adept at handling censored data—instances where the event has not occurred for some participants by the end of the study or remains unobserved. To address these unique challenges, specialized techniques like the Kaplan-Meier estimator, log-rank test, and Cox...
01:29Bias in Epidemiological Studies
Biases can arise at various stages of research, from study design and data collection to analysis and interpretation. Recognizing and addressing these biases is essential to ensure the validity and reliability of epidemiological findings.Broadly speaking, biases in epidemiology fall into three main categories: selection bias, information bias, and confounding. A more detailed description of possible biases is: