Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T09:41:31.201Z Has data issue: false hasContentIssue false

Accounting for Skewed or One-Sided Measurement Error in the Dependent Variable

Published online by Cambridge University Press:  14 January 2021

Daniel L. Millimet
Affiliation:
Department of Economics, Southern Methodist University, Dallas, TX 75275-0496, USAInstitute of Labor Economics, 53113 Bonn, Germany
Christopher F. Parmeter*
Affiliation:
Miami Herbert Business School, University of Miami, Coral Gables, FL 33146, USA. Email: cparmeter@bus.miami.edu
*
Corresponding author Christopher F. Parmeter
Rights & Permissions [Opens in a new window]

Abstract

While classical measurement error in the dependent variable in a linear regression framework results only in a loss of precision, nonclassical measurement error can lead to estimates, which are biased and inference which lacks power. Here, we consider a particular type of nonclassical measurement error: skewed errors. Unfortunately, skewed measurement error is likely to be a relatively common feature of many outcomes of interest in political science research. This study highlights the bias that can result even from relatively “small” amounts of skewed measurement error, particularly, if the measurement error is heteroskedastic. We also assess potential solutions to this problem, focusing on the stochastic frontier model and Nonlinear Least Squares. Simulations and three replications highlight the importance of thinking carefully about skewed measurement error as well as appropriate solutions.

Information

Type
Article
Copyright
© The Author(s) 2021. Published by Cambridge University Press on behalf of the Society for Political Methodology
Figure 0

Table 1Determinants of killings by Nepalese Maoists: 1996–2003.

Figure 1

Table 2Determinants of killings by Nepalese Maoists: 1996–2003.

Figure 2

Table 3Reported and predicted killings by Nepalese Maoists: 1996–2003.

Figure 3

Figure 1Distribution of observed and predicted killings. Notes: Predictions in (a) and (c) based on estimates in columns 5, 6, and 8 in Table 1, with predicted killings aggregated at the village or district level. Predictions in (b) and (d) based on estimates in columns 5, 6, and 8 in Table 2, with predicted killings aggregated at the village or district level. Kernel densities obtained using an Epanechnikov kernel and Silverman’s (1986) rule-of-thumb bandwidth. See text for further details.

Figure 4

Table 4Determinants of crime rates in Argentina: 1958–1962 cohorts

Figure 5

Table 5Determinants of per capita carbon dioxide emissions.

Supplementary material: Link

Millimet and Parmeter Dataset

Link
Supplementary material: PDF

Millimet and Parmeter supplementary material

Appendix 1

Download Millimet and Parmeter supplementary material(PDF)
PDF 394.6 KB
Supplementary material: PDF

Millimet and Parmeter supplementary material

Appendix 2

Download Millimet and Parmeter supplementary material(PDF)
PDF 357.3 KB