Hostname: page-component-7b7567c8c-7pwfn Total loading time: 0 Render date: 2026-11-16T18:45:23.525Z Has data issue: false hasContentIssue false

A Comparison of Three Approaches to Handling Incomplete State-Level Data

Published online by Cambridge University Press:  25 January 2021

J. Scott Granberg-Rademacker*
Affiliation:
Minnesota State University, Mankato

Abstract

This article compares three approaches to handling missing data at the state level under three distinct conditions. Using Monte Carlo simulation experiments, I compare the results from a linear model using listwise deletion (LD), Markov Chain Monte Carlo with the Gibbs sampler algorithm (MCMC), and multiple imputation by chained equations (MICE) as approaches to dealing with different severity levels of missing data: missing completely at random (MCAR), missing at random (MAR), and nonignorable missingness (NI). I compare the results from each of these approaches under each condition for missing data to the results from the fully observed dataset. I conclude that the MICE algorithm performs best under most missing data conditions, MCMC provides the most stable parameter estimates across the missing data conditions (but often produced estimates that were moderately biased), and LD performs worst under most missing data conditions. I conclude with recommendations for handling missing data in state-level analysis.

Information

Type
The Practical Researcher
Copyright
Copyright © The American Political Science Association, 2007

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable