<abstract xmlns="http://www.w3.org/1999/xhtml"><p>Misspecification effects (<italic>meffs</italic>) measure the effect on the sampling variance of an estimator of incorrect specification of both the sampling scheme and the model considered. We assess the effect of various features of complex sampling schemes on the inferences drawn from models for panel data using <italic>meffs</italic>. Many longitudinal social survey designs employ multistage sampling, leading to some clustering, which tends to lead to <italic>meffs</italic> greater than unity. An empirical study using data from the British Household Panel Survey is conducted, and a simulation study is performed. Our results suggest that clustering impacts are stronger for longitudinal studies than for cross-sectional studies, and that <italic>meffs</italic> for the regression coefficients increase with the number of waves analysed. Hence, estimated standard errors in the analysis of panel data can be misleading if any clustering is ignored.</p></abstract>

Misspecification effects (meffs) measure the effect on the sampling variance of an estimator of incorrect specification of both the sampling scheme and the model considered. We assess the effect of various features of complex sampling schemes on the inferences drawn from models for panel data using meffs. Many longitudinal social survey designs employ multistage sampling, leading to some clustering, which tends to lead to meffs greater than unity. An empirical study using data from the British Household Panel Survey is conducted, and a simulation study is performed. Our results suggest that clustering impacts are stronger for longitudinal studies than for cross-sectional studies, and that meffs for the regression coefficients increase with the number of waves analysed. Hence, estimated standard errors in the analysis of panel data can be misleading if any clustering is ignored.

Misspecification Effects in the Analysis of Panel Data

Departamento de Estatística e Programa de Pós-Graduação em Economia, Universidade Federal de Juiz de Fora (UFJF), Rua José Lourenço Kelmer, s/n, Campus Universitário, Bairro São Pedro, 36036-900, Juiz de Fora, MG, Brazil.

Southampton Statistical Sciences Research Institute (S3RI), University of Southampton, Southampton, SO17 1BJ, United Kingdom.

Instituto Universitário de Lisboa (ISCTE-IUL), Business Research Unit, Av. Forças Armadas, 1649-026, Lisbon, Portugal.

Journal of Official Statistics

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

{"article-title":"Misspecification Effects in the Analysis of Panel Data"}

Misspecification effects (meffs) measure the effect on the sampling variance of an estimator of incorrect specification of both the sampling scheme and the...

Misspecification Effects in the Analysis of Panel Data

Published Online: May 28, 2016

Page range: 487 - 505

Received: Jul 01, 2014

Accepted: Jan 01, 2016

DOI: https://doi.org/10.1515/jos-2016-0025

Keywords
Longitudinal survey, sampling variance, multistage sampling, stratification, weighting

© 2016 Marcel de Toledo Vieira et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Misspecification Effects in the Analysis of Panel Data

Published Online: May 28, 2016

Page range: 487 - 505

Received: Jul 01, 2014

Accepted: Jan 01, 2016

DOI: https://doi.org/10.1515/jos-2016-0025

KeywordsLongitudinal survey, sampling variance, multistage sampling, stratification, weighting

© 2016 Marcel de Toledo Vieira et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Keywords
Longitudinal survey, sampling variance, multistage sampling, stratification, weighting