Comparison of performances of heteroskedasticity tests under measurement error

Selin Alıca; Şenay Açıkgöz; Rukiye Dağalp; Şahika Gökmen

Makale detayları

Comparison of performances of heteroskedasticity tests under measurement error

Yazarlar:

Selin Alıca | Şenay Açıkgöz | Rukiye Dağalp | Şahika Gökmen

Yayıncı:

Ankara University

Ana Başlıklar:

Statistical Theory | Statistics (Other)

Dergi:

Communications Faculty of Sciences University of Ankara Series A1 Mathematics and Statistics Vol.74 Issue 2, pp.333-345

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348173

pISSN:

1303-5991

Anahtar kelimeler:

Özet:

 While measurement error has an impact on the unbiasedness of the ordinary least squares OLS estimator, the heteroskedastic error term causes inefficient OLS estimators and biased variance estimates. Although the econometric literature has answers to these two fundamental concerns, such as applying measurement error correction methods and heteroskedasticity-robust standard errors, they do not directly address testing heteroskedasticity. This paper investigates the power of the most commonly used heteroskedasticity tests in the presence of error-in-variables. Monte Carlo simulations under different heteroscedasticity forms and sample sizes show that since measurement error inflates the variance of the explanatory variable and the response variable, heteroskedasticity tests lose their power in detecting heteroskedasticity. Simulations also show that the Glejser test is the most powerful one while the White test is weak, and the other tests lie in between them.

 While measurement error has an impact on the unbiasedness of the ordinary least squares OLS estimator, the heteroskedastic error term causes inefficient OLS estimators and biased variance estimates. Although the econometric literature has answers to these two fundamental concerns, such as applying measurement error correction methods and heteroskedasticity-robust standard errors, they do not directly address testing heteroskedasticity. This paper investigates the power of the most commonly used heteroskedasticity tests in the presence of error-in-variables. Monte Carlo simulations under different heteroscedasticity forms and sample sizes show that since measurement error inflates the variance of the explanatory variable and the response variable, heteroskedasticity tests lose their power in detecting heteroskedasticity. Simulations also show that the Glejser test is the most powerful one while the White test is weak, and the other tests lie in between them.

 While measurement error has an impact on the unbiasedness of the ordinary least squares OLS estimator, the heteroskedastic error term causes inefficient OLS estimators and biased variance estimates. Although the econometric literature has answers to these two fundamental concerns, such as applying measurement error correction methods and heteroskedasticity-robust standard errors, they do not directly address testing heteroskedasticity. This paper investigates the power of the most commonly used heteroskedasticity tests in the presence of error-in-variables. Monte Carlo simulations under different heteroscedasticity forms and sample sizes show that since measurement error inflates the variance of the explanatory variable and the response variable, heteroskedasticity tests lose their power in detecting heteroskedasticity. Simulations also show that the Glejser test is the most powerful one while the White test is weak, and the other tests lie in between them.

 While measurement error has an impact on the unbiasedness of the ordinary least squares OLS estimator, the heteroskedastic error term causes inefficient OLS estimators and biased variance estimates. Although the econometric literature has answers to these two fundamental concerns, such as applying measurement error correction methods and heteroskedasticity-robust standard errors, they do not directly address testing heteroskedasticity. This paper investigates the power of the most commonly used heteroskedasticity tests in the presence of error-in-variables. Monte Carlo simulations under different heteroscedasticity forms and sample sizes show that since measurement error inflates the variance of the explanatory variable and the response variable, heteroskedasticity tests lose their power in detecting heteroskedasticity. Simulations also show that the Glejser test is the most powerful one while the White test is weak, and the other tests lie in between them.

Makale detayları

Comparison of performances of heteroskedasticity tests under measurement error

Yazarlar:

Selin Alıca | Şenay Açıkgöz | Rukiye Dağalp | Şahika Gökmen

Yayıncı:

Ankara University

Ana Başlıklar:

Statistical Theory | Statistics (Other)

Dergi:

Communications Faculty of Sciences University of Ankara Series A1 Mathematics and Statistics Vol.74 Issue 2, pp.333-345

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348173

pISSN:

1303-5991

Heteroskedasticity

measurement error

classical linear regression model

errors in variables

Sayı

Makale detayları

Comparison of performances of heteroskedasticity tests under measurement error

Yazarlar: Selin Alıca | Şenay Açıkgöz | Rukiye Dağalp | Şahika Gökmen

Yayıncı: Ankara University

Ana Başlıklar: Statistical Theory | Statistics (Other)

Dergi: Communications Faculty of Sciences University of Ankara Series A1 Mathematics and Statistics Vol.74 Issue 2, pp.333-345

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: İngilizce

Document ID (Org No.): 2001348173

pISSN: 1303-5991

Heteroskedasticity

measurement error

classical linear regression model

errors in variables

Sayı

Yazarlar:

Selin Alıca | Şenay Açıkgöz | Rukiye Dağalp | Şahika Gökmen

Yayıncı:

Ankara University

Ana Başlıklar:

Statistical Theory | Statistics (Other)

Dergi:

Communications Faculty of Sciences University of Ankara Series A1 Mathematics and Statistics Vol.74 Issue 2, pp.333-345

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001348173

pISSN:

1303-5991